SCROLL NEWS / DISCOVERY
Search headlines
Find the stories shaping the conversation.
Results for " scaling " 184 found 🔀 AI-powered Shuffle
AMD FSR 4.1.2 Now Runs On Intel Arc XMX Cores, Delivering Better Image Quality Than XeSS Upscaling With Similar Performance
Modding community brings AMD FSR 4.1.2 to Intel Arc GPUs, showing superior image quality and efficiency with upscaling times as low as 0.59 ms.
Banijay's Frederic Balmary on Expanding English-Language Drama Output
Banijay's chief business officer Frederic Balmary on scaling up english-language drama output: 'It's clear that Netflix is looking for more U.S. hits'
There’s a Reason That Netflix’s Ad Business Has Underwhelmed
Somewhat lost in the company’s slew of firings and engagement warnings over the past few weeks was the fact that Netflix is scaling back its ad forecast for 2026 from $3 billion to $2.7 billion. That’s despite the fact that: CTV is growing by 14-15% this year This year saw a monster upfront Netflix is the biggest streamer out there this side of YouTube, and was the platform that for years brands were supposedly dying to be a part of In other words, that’s an alarming pull-back from Netflix. A Wall Street analyst I spoke with last week was blunt, calling Netflix’s ad business “disappointing.” To me, a red flag on the ad front is that Netflix has never revealed how many ad-supported subscribers it has. Because if I know anything about how big companies work, if they have a positive number that makes them look really good - they tend to share it. (How many subs does go90 have right now? We’re not saying just yet. How are ads going on ChatGPT? Oh, a billion run rate - I see). Yes, to date Netflix has shared some fuzzy ad numbers, such as the idea that the company’s ad tier ‘reaches’ 250 million users globally.’ Here’s why that figure gives me pause: a)they lead with global, and b) they don’t explain how they are calculated. (Are they based on average people in each room? Surveys of some kind?). Regardless, some insiders say that Netflix’s ad tier is scaling nicely - others not so much. Either way, one of Netflix’s biggest challenge on the ad front appears to have been its starting point. Remember, when Netflix launched ads in 2022, it had 73 million subs in the US and Canada. That’s 73 million people who were used to Netflix with zero ads. And when the ad product rolled out, the company clearly was going after new users - a limited pool of potential subscribers by default. That focus seems to have changed over time, as Netflix has tried to convert longtime subscribers to the ad tier. Historically Netflix subscribers are very sticky. But is the company’s low churn a problem? Not necessarily, according to Antenna data. Rather, Netflix’s consumer loyalty is a bit of a double-edged sword. “Netflix is the one streamer which actually has seen substantial “trade downs” but has much lower ‘win backs,’ said Antenna founder Jonathan Carson (win-backs being people who were on the ad-free plan, cancelled, then rejoined on the ad plan). “Netflix has lower churn and so has less opportunity to win back old subs with a lower cost plan option.” This all might explain why Netflix has been so slow to roll out all of these table stakes targeting options until just this week. It likely didn’t have enough users to carve up into various targets. etc. However, what really hit Netflix’s ad business was Amazon’s move a few years ago to turn on advertising for Prime Video by default. Suddenly the market was awash in premium inventory, and Netflix was playing catch-up on multiple fronts. Thus, the streaming giant appears to still be figuring out if if wants lead in ‘programmatic TV,’ or premium ‘upfront’ TV sales. “The CTV market now sorts into three businesses, and Netflix is not the leader of any of them,” wrote analyst Jack Myers. Ouch. So how does Netflix ratchet up its ads business? Get the next NFL package You’re gonna have to wait until 2030 unless one of the bigger media companies like Skydance opts out, but the easiest (not cheapest) way for Netflix to become a huge TV ads contender is to get one of the regular NFL packages, ideally Sunday afternoons. That way it can sell ads to its whole U.S. subs base overnight. Go FAST This one is risky, because to program a compelling FAST channel, you’d likely have to take some older Netflix exclusive shows and make them available for free to non-subscribers. But FASTs, for better or worse, are ad-inventory machines. Go full UGC I’m not sure I see this happening. There is something already off about Netflix carrying podcasts and shows for creators. The experience sticks out in what is such a produced, and generally non-communal environment. But Netflix could open up its platform by allowing creators to upload videos without having to be approved by the service. But that opens a whole other can of worms (policing brand safety, copyright infringement, overall quality). Make a huge RMN Deal I was going to suggest something crazy, such as ‘instead of buying Warner Brothers, Netflix should buy Target. But the company doesn’t want to be in the brick and mortar business. Yet for Netflix’s ad business to compete in an Amazon world, it needs to be able to connect ad exposures to shopping. What if, instead of letting Amazon target Netflix viewers with its DSP, Netflix inked a deal with Target, or even took control of Roundel, and created its version of the Amazon full-funnel for brands? Price us out. I hope they don’t do this, but Netflix’s price increases - while significant - haven’t been enough to push long-term customers out of the ad-free tier. But if they made those prices so prohibitive, it could push large swaths of consumers to trade down. Then again, that could also dent engagement - as binging becomes much less appealing once you bring ads into the equation. Or, they could just launch a bunch of hit shows and maybe everyone will quickly forget all this angst. More from Amazon unBoxed A few weeks ago, I recorded two podcasts live at Amazon’s unBoxed event in San Francisco. I was able to go really deep with several brand partners about how Amazon’s various data and ad products work - particularly the under appreciated Amazon Marketing Cloud. First, I talked with Laura Abadias Velilla, of the Amazon specialty agency Amerge, along with Amazon’s own Charlotte Maines, Vice President, Devices Content & Advertising, about how the two partners helped a very specific printer company Epson launch into a brand new category - high end projectors. It’s a great example of how brands connect audience data across platforms to make budget deployment decisions. Next, I chatted with Monks VP Head of Digital Investment for North America Tom Burchill about how the agency actually brought in data from Linkedin to help target B2B customers across various Amazon properties for the brand Quicken. Check it out: Check these out, or just listen:
World News in Brief: Aid for Ethiopia, Abyei drone strikes condemned, extreme heat and newborn health
The World Health Organization (WHO) is scaling up emergency health response operations across Afar, Amhara and Tigray in Ethiopia to address growing humanitarian and health needs resulting from ongoing hostilities.
Xbox follows in PlayStation's footsteps, introducing AMD FSR 4.1, targeting better graphics and performance on Series X/S
Xbox Series X and Series S are getting a new AI upscaling technology comparable to PlayStation's QSSR.
Miami-Dade Small Business Program: New opportunities
Getting your Trinity Audio player ready... For a small business owner, winning a government contract can be a significant milestone — obtaining steady revenue, scaling up your business, and achieving long-term growth. In Miami-Dade County, an important first step for many businesses is becoming certified through the county’s Small Business Development programs. Today, Miami-Dade County […]
Microsoft is now scaling back Copilot in Windows 11 amid Microslop backlash, pulling 8 AI actions from File Explorer and calling it “simplifying”
Microsoft says it's retiring "Copilot actions" from the File Explorer context menu, which it recently began rolling out.
AMD is bringing FSR 4.1 to Xbox consoles and handhelds
AMD’s latest upscaling tech is also coming to laptops, 6000-series GPUs, and handhelds.
AMD and Microsoft Bring FSR 4.1 to Xbox Series X|S, FSR Diamond Teased for Project Helix
AMD and Microsoft have confirmed that FSR Upscaling 4.1 is coming to the Xbox Series X and Series S. The news came in an AMD video presentation led by Jack Huynh, joined by Xbox CTO Scott Van Vliet and ASUS. The ROG Xbox Ally handhelds are getting it too, and it looks like they will be first in...
Revenue And Earnings Growth Could Be A Game Changer For agilon health (AGL)
agilon health has reported strong annual revenue and earnings growth from its model of partnering with primary care physicians to deliver value-based care for Medicare patients through long-term, globally capitated contracts. The combination of accelerating financial performance and the scaling of its value-based primary care network indicates that agilon health is gaining share where payers and physicians seek partners willing to take on population-level risk. We will now look at how agilon...
Core Growth Group Updates HVAC Business Scaling Guide With Sequenced Growth Plan
The revised guide, updated October 1, 2026, sets out a four-step scaling sequence, a 24- to 36-month timeline, and the operational changes that ...
Scaling perinatal mental health care: Scientists assess home visiting programs in Midwest
Can reaching out to new mothers right at the outset of parenting set them on the right path? A new study published in BMC Health Services Research assesses home health visitation programs, which provide ...
Scaling the Rocks: Ding dong the Dodgers are dead
Colorado Rockies news and links for Friday, October 16, 2026
Nicklaus Children’s Health System Appoints Brynne McNulty Rojas to Board of Directors - Florida Hospital News and Healthcare Report
October 14, 2026 – Nicklaus Children’s Health System has appointed entrepreneur and investor Brynne McNulty Rojas to its Board of Directors. McNulty Rojas brings expertise in technology scaling, finance, real […]
Healf rides youth health kick to top UK’s fastest-growing companies list
The wellbeing start-up, which has its roots in Manchester, is scaling quickly
Teradata Brings Elastic Python Compute to AI Studio, Unifying Data, Governance, and AI
New distributed Python capability accelerates large-scale AI operations while keeping data and compute within the Teradata platform SAN DIEGO, Oct. 14, 2026 — Teradata today announced the general availability of Elastic Python Compute within AI Studio. The capability gives enterprise data science teams a managed, distributed Python environment for scaling AI compute and machine learning […]
Confido Health nabs $12.5M to scale agentic AI across PE-backed medical groups
Confido Health is expanding beyond AI voice agents to build out an agentic AI operating system that can execute administrative workflows across a medical practice. | The company secured a $12.5 million series A extension following its $10 million series A round announced in September to continue scaling an AI platform for end-to-end healthcare operations.
Owen Tripp, Included Health CEO: A Fortt Knox Update
Included Health CEO Owen Tripp told Jon Fortt that after its Firefly Health acquisition, the company is scaling the alternative health plans it launched in 2026. Tripp described a shift to performance-based contracts, where it shares in savings it generates rather than taking a flat admin fee. On its plan, 76% of the highest-cost members engage voluntarily, and he expects the market to shift within five years.
Business priorities for mobilising and scaling finance for biodiversity - ICC
COP17 provides a window to turn global biodiversity finance commitments into investment at scale. With at least US$200 billion a year needed for biodiversity by 2030, this ICC policy paper sets out the frameworks, financing solutions and market conditions needed to reduce investment risks and channel capital towards measurable outcomes for nature.
Building a science of biodiversity recovery
Nature Reviews Biodiversity - Biodiversity science has built powerful systems for detecting decline, but must now pivot towards defining, explaining and scaling recovery. The 17th Conference of the...
Dealroom.co | Europe's AI-and-Science Edge: Raffi Kamber on Turning Labs into Companies
Key takeaways from Scaling Europe interview with Singular co-fou…
Amazon Small Business Event: Empowering the Next Generation and Scaling Service-Based Entrepreneurs
Join Hera Hub and Amazon for a hands-on workshop designed to help aspiring entrepreneurs and established service-based business owners bring new products to life. Amazon experts and small business owners will provide practical guidance, useful tools, and actionable next steps to launch a product and build a business with Amazon. This event is free but space is limited to 60 participants and will fill up. The event includes a light breakfast and lunch. Participants will choose from two specialized tracks based on their goals and stage of business. TRACK 1: NEXT GENERATION From Product Idea to Online Store Designed for aspiring young entrepreneurs, from junior high students through college and beyond, who have a product idea or want to explore the world of e-commerce. Amazon small business owners and Amazon experts will review topics such as: • How to develop and refine a product idea • How to set up and sell products on Amazon • What to consider when pricing, marketing, and fulfilling orders • Practical next steps for moving from idea to launch Whether participants already have a product in mind or are simply curious about entrepreneurship, this track will help them see what is possible and give them a roadmap to get started. TRACK 2: SERVICE EXPANSION Turn Your Expertise Into a Product Designed for consultants, coaches, speakers, creatives, and other service-based entrepreneurs who want to expand beyond trading time for money. Participants will explore how to transform their knowledge and expertise into tangible products such as: • Books and workbooks • Journals and planners • Card decks • Educational tools and resources Amazon small business owners will share how they identified viable product opportunities and created a product that can generate revenue beyond their one-on-one services. An Amazon expert will also review self-publishing and other ways to build a business with Amazon. By the end of the workshop, participants will leave with greater clarity, practical guidance, and a tangible next step toward launching or expanding their product-based business. The event includes a light breakfast and lunch. This event is free but space is limited to 60 participants and will fill up.
Scaling the Rocks: CarGo joins elite Colorado club
Colorado Rockies news and links for Friday, October 9, 2026
Climate hopes shift from politics to the courts
The political fight against global warming may be losing steam, but the legal battle is heating up. On the political front, from Europe to the United States, governments have been scaling back climate...
A Business and Regulatory Navigator to Accelerate Electrification
Demand for electrification is moving faster than the system it connects to. Electric vehicles, heat pumps and electrified industrial processes are scaling1, and the the United Nations Climate Change Conference (COP31) Presidency has set a global target of 35% of final energy demand met by electricity by 2035, up from around 21% today.2 What now
South Korea considers scaling back athlete military service exemption, sports minister says
Three footballers apologised this week for appearing to celebrate avoiding conscription after winning Asian Games gold.
Why advanced diagnostics are essential for health equity in Asia
The Asia-Pacific region faces a growing diagnostic gap. Discover how scaling advanced diagnostics can improve healthcare equity, efficiency and patient outcomes across the region.
Scaling the Rocks: The Rockies are looking ahead
Colorado Rockies news and links for Tuesday, October 6, 2026
Sony introduces AI upscaling to base PS5; Marvel's Wolverine and Ghost of Yōtei are first games to implement new tech
PlayStation Spectral Super Resolution (PSSR) remains exclusive to the PS5 Pro, which SIE says is the "gold standard"
Some mental health providers hoping to serve Lane County's Oregon Health Plan members say they’re facing barriers
Aspire is one of a few healthcare providers in Lane County that told KLCC they want to – but haven’t been able to – get a contract with Lane County's Medicaid insurer Trillium. That’s left Aspire without access to its target population, and other providers considering scaling back their services.
PS5 gets "AI-guided upscaling" a bit like PS5 Pro's, and it's available now in Marvel's Wolverine and Ghost of Yotei
AI upscaling was previously exclusive to PS5 Pro, but base PS5s are getting a watered down version, and it's available now.
Sony Brings AI Upscaling To The Base PS5
Marvel’s Wolverine and Ghost of Yōtei are the first games to support Quick Spectral Super Resolution.
Sony brings AI graphics upscaling to the regular PS5
It will be available today in two games.
Hands-On With QSSR: AI-Driven Upscaling for the Standard PS5
Plus: further details from Sony's graphics R&D team.
Sony announces QSSR AI upscaling tech for base PS5 in Marvel's Wolverine and Ghost of Yōtei
The PS5 Pro is no longer the sole domain of AI upscaling tech on console.
Gears of War: E-Day Performance Benchmark Review
Gears of War: E-Day is a prequel set 14 years before the original game. Built on Unreal Engine 5 with RTX Mega Geometry, DLSS 4.5, FSR 4, XeSS, and ray tracing, we examine graphics quality, VRAM usage, upscaling options, and how it performs across a wide range of modern GPUs.
Global Health & WASH: October 2026 Funding Opportunities (26 new opportunities)
41 opportunities on the Global Health & WASH page this cycle, 26 of them new. The health money on this page is mostly research money, and much of it is aimed at people early in their careers. Funder after funder, from the TANGO2 Research Foundation’s Research Pathways grants to the South African Medical Research Council’s Early Investigators Programme to the National Ataxia Foundation and Alex’s Lemonade Stand Foundation, is funding early-career investigators and the institutions that host them, with a clear tilt toward researchers based in or from the Global South, from AMBSO in Uganda to Kenya’s CONNECT-VAX network to Ersilia’s AI Incubator for African drug discovery. The money is organized by disease as much as by need: narrow foundations, the Parkinson’s Foundation, NTAP for neurofibromatosis, the RUNX1 Research Program for an inherited blood disorder, each run their own grant, so the question for an applicant is less which health problem you work on than which foundation owns your disease. The largest European cheques go to consortia only, with the European Innovation Council putting up to EUR €4 million behind a single healthy-ageing project and UNITE up to EUR €1 million behind each cross-border digital health team. Water sits apart, almost entirely in applied utility research: the Water Research Foundation’s cluster of USD $250,000 to $500,000 projects, each built around a utility problem from sewer laterals to reuse treatment. And the biggest single pot is public: HRSA is renewing community health center service areas with awards of up to USD $10 million each. This post is for paid subscribers. This helps support the time and effort it takes to curate and organize these opportunities. Subscribe To keep this accessible to everyone who needs it, we’re happy to offer pay what you can rates. You can find more details here. Open Opportunities 2027 Building the Field of Health Advocacy Grant, Georgia Health Initiative. *New!* *Closing soon!* Georgia Health Initiative, a nonprofit private foundation focused on health equity, is opening its 2027 Building the Field of Health Advocacy Grant to strengthen the organizations that push for healthier systems across the state. The Initiative treats advocacy as infrastructure, backing one-year projects that range from urgent responses to shifting federal policy to longer arc campaigns on housing, behavioral health, care access, and food security. It wants coalitions that elevate community voice, apply an equity lens, and arrive ready to deploy concrete strategies, and it deliberately seeks variety in issues, communities, and organizational size. The optional webinar and required lobbying training signal a funder investing in a durable field, not one grant cycle. The strongest fits are Georgia nonprofits and public agencies ready to turn advocacy readiness into systems change. Geographies: United States. Who can apply: Tax-exempt 501(c)(3) public charities, state or local government agencies, or coalitions/collaboratives using a 501(c)(3) fiscal sponsor, that work in Georgia on behalf of Georgians and align with the mission, vision, and values of Georgia Health Initiative. Only one application per organization, though an organization may be listed as a partner or subcontractor on multiple applications. Funding amount: Organizations can apply for up to USD $80,000 per project; the Initiative anticipates investing a total of USD $1.36 million. Targeted Sectors / SDGs: Governance, Policy & Advocacy · SDG3; SDG10. Deadline: October 2, 2026. Learn more and apply here. TDD Research Pathways Grant, TANGO2 Research Foundation. *New!* *Closing soon!* The TANGO2 Research Foundation, working with a founding gift from Tango2UK, is opening its TDD Research Pathways Grant to fund early-career investigators anywhere in the world who study TANGO2 Deficiency Disorder. The foundation is using this money to seed pilot work and to build a research community around a rare disorder that few labs currently study. It wants to back emerging scientists, clinicians, and trainees who will stay with this field, not one-off projects, which is why it requires a committed mentor and real institutional support and asks each applicant for a plan for future contributions. The design rewards promise and persistence over track record. The strongest fits are students, residents, postdoctoral fellows, and new investigators ready to commit to TANGO2 research. Geographies: Global. Who can apply: Early-career investigators worldwide, including undergraduate, graduate, and medical trainees, postdoctoral fellows, residents, and clinical or research fellows. Early-career investigators must hold a PhD, MD, or equivalent doctoral degree and be within 10 years of that degree or within 5 years of initiating rare disease research, engaged in postdoctoral research at an academic institution, teaching hospital, or nonprofit research institute. Currently enrolled trainees and transitional PhD candidates beginning a postdoctoral appointment at a nonprofit institution by July 1, 2026 are also eligible. All applicants must identify a qualified mentor with relevant TDD or rare disease research experience and demonstrate institutional support. Funding amount: Up to USD $15,000. Targeted Sectors / SDGs: Innovation, Science & Technology. Deadline: October 2, 2026. Learn more and apply here. Call for Expression of Interest (EOI): AMBSO Impact Research Initiative, Africa Medical and Behavioral Sciences Organization (AMBSO). *Closing soon!* The Africa Medical and Behavioral Sciences Organization is inviting expressions of interest for its Impact Research Initiative, which funds early-stage pilot studies and proof of concept work on priority health challenges. AMBSO is a Ugandan research organization backing bold ideas that can grow into larger studies, and its tracks span antimicrobial resistance, epidemic preparedness, artificial intelligence for diagnostics, HIV prevention, climate and health, cancer, maternal and child health, and non-communicable diseases. The small award size and direct-cost-only rule point to a funder that wants evidence rather than overhead, and that favors projects able to attract follow-on money or use AMBSO’s own research infrastructure. Eligibility runs worldwide, welcoming researchers and institutions inside and outside Africa. The strongest fits are researchers with a feasible pilot that generates policy-ready evidence. Geographies: Global. Who can apply: Open worldwide to researchers, academic institutions, and research organizations whose work aligns with generating high-impact health evidence. Independent researchers are eligible if affiliated with an academic or research institution. Doctoral students are encouraged to apply as part of a research team led by a faculty Principal Investigator. Community-based organizations and other eligible entities may apply. Projects may be implemented inside or outside Uganda. Funding amount: Up to USD $10,000 per project, with no minimum funding threshold. Targeted Sectors / SDGs: SDG3. Deadline: October 2, 2026. Learn more and apply here. Entrepreneurs for Resilience 2027 (E4R27), Swiss Re Foundation. *Closing soon!* Swiss Re Foundation is opening the tenth edition of Entrepreneurs for Resilience, which backs healthcare social enterprises expanding affordable, quality primary care for low-income communities across low- and middle-income countries. The Foundation is using catalytic, blended funding plus multi-year advisory to take risk out of scaling, helping proven models reach more people sustainably rather than stay stuck at pilot size. It looks to back brick-and-click enterprises that pair physical care with digital tools and build affordability through savings, payments, or microinsurance, ideally integrated with public health systems. Requiring earned revenue, real customer scale, and gender-inclusive teams signals that this is growth capital for commercially serious ventures, not seed money. The strongest fits are established, revenue-generating healthcare enterprises ready to scale access for underserved populations. Geographies: Global. Who can apply: For-profit and hybrid healthcare social enterprises serving low-income and underserved populations in low- and middle-income countries, using combined physical and digital (brick-and-click) healthcare delivery models. Applicants must demonstrate measurable social impact, generate at least 50% of income through earned revenue, have been established no later than December 2023, already serve at least 50,000 people with potential to scale, and have at least 30% low-income customers. Preference for gender-inclusive enterprises with at least one female founder or at least 30% women in management. Funding amount: Pilot funding of up to USD $45,000 for the 6-8 shortlisted social enterprises; up to USD $1 million in blended funding (up to USD $600,000 unconditional plus up to USD $400,000 conditional) shared among 3-4 finalists over a two-year implementation period; total programme funding approximately USD $1.3 million. Targeted Sectors / SDGs: Innovation, Science & Technology · SDG3; SDG5. Deadline: October 4, 2026. Learn more and apply here. Horizon HealthTech 2026, IIITD Innovation & Incubation Center. *Closing soon!* IIITD Innovation and Incubation Center is opening Horizon HealthTech 2026, a nine-month MedTech incubation cohort run under the aegis of BIRAC EDGE for founders building medical devices, diagnostics, surgical and interventional tools, or remote monitoring. The center is backing teams that already hold a validated proof of concept and want to reach a tested prototype, pairing each founder with a named expert and clinical validation through a hospital partner. What they seek to back is Indian health technology that can survive real clinical scrutiny and regulatory review, not slideware. The structure, with lab access, CDSCO guidance, IP support, and investor connects, signals a program built to make startups ready for follow-on capital. The strongest fits are early-stage MedTech teams with defensible IP and a working path from bench to bedside. Geographies: India. Who can apply: Early-stage teams/startups with a validated idea/proof of concept (TRL 3 onwards) working on medical devices, diagnostics, surgical/interventional technologies, or remote monitoring, holding IP or with potential to generate IP. Funding amount: No cash award amount is stated on the official program page. Benefits are in-kind: prototyping lab and equipment access, clinical/hospital validation (Max Healthcare), expert mentorship, CDSCO regulatory guidance, IP filing support, workspace, cloud credits and tools, and investor connects. Run under the aegis of BIRAC EDGE and builds a path to follow-on Government/VC/Angel funding. Targeted Sectors / SDGs: Innovation, Science & Technology · SDG3; SDG9. Deadline: October 5, 2026. Learn more and apply here. OpenAI AI and Teen Development Research Grant Program, OpenAI. *New!* *Closing soon!* OpenAI is opening its AI and Teen Development Research Grant Program to fund independent studies of how generative AI shapes the lives of young people ages 13 to 17. OpenAI wants a stronger evidence base on how teens actually use these tools and what that use does to their social and emotional development, so that product teams, policymakers, and regulators can act on findings rather than assumptions. It seeks work on emotion regulation, relationships, belonging, and the safeguards that keep teens safer, welcoming interdisciplinary teams and any methodology. That it will fund research whose conclusions may not favor AI providers, and prioritizes studies touching ChatGPT, signals a genuine appetite for credible outside scrutiny. The strongest fits are research groups grounded in adolescent development, well-being, and AI safety. Geographies: Global. Who can apply: Researchers aged 18 or older who are affiliated with a research institution or organization, and/or have significant experience relevant to child or adolescent development, well-being, human-computer interaction, AI impacts, or a closely related field. Multiple institutions may collaborate. OpenAI is seeking to fund research rather than for-profit initiatives and will not prioritize for-profit organizations. Funding amount: Individual research grants up to USD $1 million, totaling up to USD $5 million; maximum 10% of each grant for overhead or indirect costs. Targeted Sectors / SDGs: Social Inclusion & Community Wellbeing; Innovation, Science & Technology. Deadline: October 6, 2026. Learn more and apply here.
UN Decade of Healthy Ageing midpoint report finds progress, calls for greater investment
New evidence in the Secretary-General's report to the 81st UN General Assembly, prepared by the World Health Organization (WHO) in coordination with the UN system, finds more countries acting on healthy ageing than ever before – but longer lives will not become healthy, dignified ones without faster, more sustained investment at country level.(NEW YORK, 28 September 2026) The Secretary-General transmitted to the 81st session of the UN General Assembly the mid-term progress report on the UN Decade of Healthy Ageing (2021–2030), marking the midpoint of the ten-year global effort to improve the lives of older people. The report was prepared by the World Health Organization (WHO) in coordination with the UN Inter-Agency Group on Ageing and with inputs from thirteen UN entities, the World Bank Group and civil society partners. It finds real progress in countries – and persistent gaps that must close before 2030.The number of people aged 60 and over passed 1 billion in 2020 and is projected to reach 2.1 billion by 2050, with the fastest growth in low- and middle-income countries. Longer lives are not automatically healthy ones: globally, the gap between life expectancy and healthy life expectancy at age 60 widened between 2000 and 2023, and an estimated 71% of excess deaths from COVID-19 between 2020 and 2023 were among people aged 60 and over. “Longer lives are one of humanity’s great achievements, but we cannot congratulate ourselves while leaving millions without health and support,” said Dr Pascale Allotey, Director, Department of Sexual, Reproductive, Maternal, Child and Adolescent Health and Ageing, WHO. “The first half of the Decade has shown what is possible. The second half must be about scaling and financing the progress to meet the needs and rights of older persons.”Real progress, unevenly sharedThe report finds that more countries than ever now have legislation against age-based discrimination – up from 44.5% in 2018 to 57.7% today – and over half now have national long-term care policies, up from 49% in 2023 to 56.7%, dedicated programmes to build age-friendly environments (54.6%), and comprehensive health and social assessments of older people (54.6%). New WHO guidance on services for long-term care and for Integrated Care for Older People (ICOPE) delivered in communities is enabling systemic change as an increasing number of countries incorporate these services within national universal health care benefit packages. A host of UN agencies and other partners support training, increasing health workforce competencies and use of digital health tools.“This report shows the Decade of Healthy Ageing is working,” said Dr Ritu Sadana, Head of the WHO Secretariat for the UN Decade of Healthy Ageing. “More countries are legislating against ageism, providing integrated care, strengthening long-term care systems, and creating age-friendly environments than ever before, in all regions and income levels. But progress is uneven and in one area, regressive: fewer countries report active multi-stakeholder forums on ageing, down from 53.1% in 2020 to 43.3% today. The second half of this Decade must turn commitments into results, in every country.”The financing gap: this Decade's binding constraintFinancing remains the binding constraint: more than a third of reporting countries (33.8%) have very little or no dedicated resources to develop long-term care, with similar shortfalls across the Decade’s other priority areas. Less than half of countries (45.4%) have legislation providing access to assistive devices (including hearing aids, spectacles or rollators), which are key to enabling older people to maintain productive, independent and dignified lives. Only 22.7% of countries can track older people’s health over time – the largest gap across the Decade’s core progress indicators. Nationally representative information is needed in each country, as policies and interventions cannot be designed as if all older adults of a given age share similar needs and capacities. The report calls for sustained investment through national budgets, country-owned plans, development cooperation and innovative financing mechanisms to fully close these gaps by 2030.“The Decade was designed as a whole-of-UN effort, and this report shows it in action,” said Amal Abou Rafeh, Chief of the Programme on Ageing Section at the UN Department of Economic and Social Affairs (UN DESA), and Permanent Co-Chair of the UN Inter-Agency Group on Ageing. “Thirteen UN entities, the World Bank group and civil society organizations contributed to this report, with WHO as the lead technical and implementing agency. That is the model of coordination the second half of the Decade will need to scale, with countries leading legislation, funding and implementation, engaging older people.”“Investing in healthy ageing is not simply a cost; it is one of the highest-return investments a government can make,” said Fiona Stewart, Global Lead for Ageing and Pensions at the World Bank Group. “Countries that invest early in integrated and long-term care see savings that outweigh the upfront cost many times over. The evidence in this report makes the economic case; what is now needed is predictable financing at country level to act on it, along with efforts to support older entrepreneurs and the silver economy.”Older people's own voicesThe report also draws on the direct experience of older people themselves, including a multilingual global survey conducted by the International Federation on Ageing that reached more than 1000 older people across 58 countries. WHO and UN partners called on Member States to use the second half of the Decade to accelerate implementation, strengthen data systems, and secure the financing needed so that the gains of longer lives reach older people everywhere. About the UN Decade of Healthy Ageing (2021–2030)Proclaimed by the UN General Assembly in December 2020, the Decade is a global collaboration bringing together governments, UN agencies, civil society, the private sector, academia and older people themselves to improve the lives of older people, their families and communities. It is coordinated by the World Health Organization together with the UN Inter-Agency Group on Ageing. The Decade's mid-term progress report was transmitted to the General Assembly by the Secretary-General, prepared by WHO, and is available in 6 languages.Full Report available: A/81/366 Implementation of the United Nations Decade of Healthy Ageing (2021–2030): Note by Secretary-General [ العربية | 中文 | English | Français | Pусский | Español
Miami-Dade County’s Small Business Certification Program Creates New Pathways to Opportunity
Getting your Trinity Audio player ready... For a small business owner, winning a government contract can be a significant milestone—obtaining steady revenue, scaling up your business , and achieving long-term growth. In Miami-Dade County, an important first step for many businesses is becoming certified through the County’s Small Business Development programs. Today, Miami-Dade County has 1,732 […]
DHA CDAO Jesus Caban on scaling AI across military health
Dr. Jesus Caban | Chief Data & Analytics Officer | Defense Health Agency
Intel Panther Lake Teardown, 18A, BSPD, GAAFET, SemiAnalysis STEEL
Panther Lake debuts the first commercial implementation of backside power delivery (BSPDN), introduces Intel’s first iteration of gate-all-around (GAA) transistors, and showcases their advanced packaging capabilities with its Foveros-S assembly. With Panther Lake, Intel’s manufacturing arc has shifted from nebulous roadmaps to shipped silicon, a significant milestone on their long road back to competitive semiconductor manufacturing. To evaluate the extent of Intel’s comeback, we tore down Panther Lake. The SemiAnalysis STEEL teardown lab breaks down advanced datacenter and AI hardware. To learn more about our pipeline or to commission a teardown, contact sales@semianalysis.com. WE’RE HIRING: Architecture, floorplan, packaging, manufacturing, and labs experts. Opportunities from system to transistor and everywhere in between. Check out our Careers page. Our teardown traces 18A from its four-sheet RibbonFETs (Intel’s marketing name for GAAFETs) and gate stacks through contacts, frontside and backside wiring, and the bonded carrier. We explain how these material and integration choices improve gate control and reduce resistance, while adding capacitance, thermal resistance, and process complexity. Our measurements put Panther Lake’s 18A compute logic and TSMC N3E GPU logic at similar logic density. However, 18A does not lead TSMC N3P, N2 or Samsung SF2 in peak density. Panther Lake’s CPU cores are incremental updates, and the high-end GPU still uses TSMC N3E. Panther Lake assembles one compute tile, one GPU tile, and one I/O tile atop a passive base tile using Intel’s Foveros-S advanced packaging. Both compute tile variants use Intel 18A. The Xe3 GPU options are a 4-core GT1 tile on Intel 3 and a larger 12-core GT2 tile on TSMC N3E. Both I/O tile variants use TSMC N6. [1], [2] Our analysis centers on the PTL-U compute tile, both the 4-core and 12-core GPU tiles, as well as the 12-lane I/O tile. In conventional chips, power and signal are routed through the same frontside metal stack towards the device frontend. Power rails consume scarce routing resources near the transistors, while tall via stacks carry VDD and VSS from the coarse upper wires to local rails. Backside power delivery (BSPD) moves the main power network behind the transistor layer, to the backside, separating it from frontside signal routing. We covered BSPD and its impacts in 2024. [3], [4], [5] Intel’s BSPD implementation, branded as “PowerVia”, routes power through dedicated backside metals to nano-TSVs, which connect those rails to local source/drain (S/D) contacts. Implementing that separation requires Intel to build the interconnect stacks from both sides of the wafer. The frontside comprises the M0-M14 signal stack, while the backside comprises the BM0-BM5 power stack. M0 and BM0 are closest to the transistors. The nano-TSVs connect the two sides, but Intel patterns and etches each via from the front after forming the contacts. A narrow via runs from the side of the contact deep into the silicon substrate. Intel then completes the frontside signal metal stack, bonds the wafer to a carrier, flips it and removes the original substrate until the buried via tips are exposed. The backside metal stack is then deposited directly on the revealed vias. The nano-TSV and backside-via profiles taper in opposite directions because Intel forms them from opposite sides of the wafer. The transistor structures form the FEOL. Local contacts and nano-TSVs connect them to the wiring. M0 begins the frontside interconnect stack. The silicon carrier remains attached above the frontside interconnects. It supports the device wafer during substrate removal and backside processing and remains part of the finished chip’s thermal path. PowerVia removes the main power distribution from the congested frontside metals, routing supply through shorter and wider backside wires. Its lateral landing still occupies area in the standard cell, so it recovers less cell area than a direct backside contact. [3] Nano-TSVs beside the logic devices carry VDD or VSS from the backside power network, while signal connections continue upward through the frontside metals. Backside Interconnects Samsung SF2 data is included for comparison to Panther Lake’s within this article. SF2 is the incumbent GAA foundry node but lacks BSPD, serving as a useful reference to evaluate 18A. A full teardown of Samsung’s S26 products, processed on SF2, will be shared soon.Nanosheet-cut EDS comparison. The PowerVia supply path runs from the backside Cu rails through Mo-lined W nano-TSVs to the local transistor contacts. In this cross section, the tapered connection spans roughly 150 nm from the contact level to BM0. The Ta liner confines Cu and promotes adhesion to the surrounding stack; the AlOₓ etch stop controls the next dielectric etch above the rail. Dielectric beneath the ribbons electrically separates the devices from the backside wiring and removes the conducting silicon body below the channel. [6] AlOₓ serves as an etchstop (ES), enabling endpointing and protecting the underlying layers. Low-volatility aluminum fluoride reaction products resist the fluorinated plasma, allowing a thin AlOₓ film to protect the metal while the surrounding low-k dielectric is removed. [6], [7]. While the BM0 and layers above the M1 lines show double AlOx layers, Our SMIC N+3 teardown showed single AlOₓ layers. SMIC uses a simpler local AlOₓ substack, while the remaining cap and etch sequence provide the required landing protection. So why double layers? The closely spaced AlOₓ doublets provide two protected endpoints in the etch sequence. Intel documents an AlOₓ/SiN/AlOₓ stack that explains the benefit. The main dielectric plasma etch stops on the first AlOₓ film; a selective wet clear opens that film; a second plasma etch removes the intermediate SiN and stops on the second AlOₓ film. The final wet clear exposes the metal landing surface. SiN is the intermediate dielectric in Intel’s published example. [8] The second stop protects the metal through a cap breakthrough. Wide openings can etch faster than narrow ones, and etch depth varies across the wafer. Metal under an early-clearing opening would otherwise be exposed while other openings still need more etching. Staged protection widens the process window and reduces metal erosion, corrosion and void formation. [8] TSMC documents AlN/AlOₓ/SiOC/AlOₓ above Cu, with AlN blocking Cu diffusion, and a simpler AlN/SiOC/AlOₓ variant that omits one AlOx film. [9] Levels with different opening sizes, aspect ratios, pattern densities and cap materials need different etch margins. A double AlOx stop is useful where another protected endpoint justifies the added processing. The extra film adds formation, selective opening and cleaning steps, plus another set of interfaces to control adhesion, moisture, and stress. These blanket films are opened through the existing via pattern, so each film does not require another lithography mask. AlOₓ adds parasitic capacitance when it replaces lower-k dielectric; two thin AlOₓ films can nevertheless contain less AlOₓ than one thick film. Total thickness, placement, and theintermediate dielectric determine the electrical cost. Deposition chemistry also changes AlOx permittivity and residual hydroxyl content, which can oxidize the underlying metal. [7], [8], [10], [11] The backside stack separates into relatively fine BM0-BM2 wiring near the devices and coarser BM3-BM5 power distribution. The largest pitch increase occurs between BM2 and BM3. BM0’s pitch closely matches the logic-row height, fitting local power delivery to the cell rows. Higher levels aggregate current through larger conductors: routing density becomes less important than low resistance and current capacity as the network approaches the package. This hierarchy provides wide power wiring for the power delivery network without consuming scarce frontside signal-routing resources. [3] The SemiAnalysis STEEL teardown lab breaks down advanced datacenter and AI hardware. To learn more about our pipeline or to commission a teardown, contact sales@semianalysis.com. WE’RE HIRING: Architecture, floorplan, packaging, manufacturing, and labs experts. Opportunities from system to transistor and everywhere in between. Check out our Careers page. Frontside Interconnects Intel 18A combines Mo-lined W contacts and nano-TSVs with a separate backside Cu power network. Samsung SF2 keeps power on the frontside, using Ti-based contact interfaces and Ta-based barriers and Co liners around Cu wiring. In 18A standard-cell rows, backside power rails supply the devices through nano-TSVs within the cells, freeing frontside routing resources. Samsung’s M0 accommodates both power and signal connections. From the device toward M0, the connection runs through a Ti-based S/D interface, W contact fill, a Mo-lined W via, and the Cu M0 wire. Mo supplies a conductive nucleation and adhesion layer for W, replacing the resistive TiN liner used in conventional W integration. This increases the effective conduction volume within the feature while retaining W fill and its established polishing, cleaning and etching processes. Intel’s Mo/W patent describes this integration tradeoff. The nano-TSV uses the same Mo-lined W construction in the backside supply path. [12] The move from TiN to Mo is an incremental change. While a full Co or Mo fill can also reduce the volume lost to liners in very small features, it requires new integration schemes that increase complexity and risk. Cu remains attractive for wider wires due to its low resistance. As wires and vias shrink, the diffusion barrier consumes an increasing fraction of their cross-section. [12], [12], [14] Intel uses Co/Ru liners at M0-M1, Co at M2-M4, and Nb at M5-M9. The lower-level liners help Cu adhere and reduce void formation during trench fills. Applied Materials’ Endura has new thermal control that facilitate wetting process, so the thin film continuity is good enough that good capillary pressure will drive Cu atoms to the via bottom without voiding. Intel’s choice to use Nb is particularly interesting. Intel’s Nb patent describes a conductive diffusion barrier intended to reduce the barrier’s contribution to resistance relative to conventional Ta-based barriers, particularly at via bottoms where all current crosses the barrier. The patent pairs Nb in coarser levels with the option of lower-cost PVD processing. [15], [16] The upper metal layers support thicker barriers formed through physical vapor deposition (PVD) despite its worse coverage and uniformity. Meanwhile, the lower metal layers require thinner barriers deposited through conformal atomic layer deposition (ALD). Co/Ru adds another material interface and requires controlled deposition and Cu fill. Changing liners and barriers by metal layer allows Intel to optimize interconnect resistance, process complexity, and reliability. [15, 16] RibbonFET, Intel’s name for its gate-all-around FETs (GAAFETs), replaces the FinFET’s vertical fins with four stacked horizontal silicon nanosheets, allowing the gate to surround the channel on every side. The path to GAAFET begins with the planar transistor. A planar MOSFET places the gate above the channel between its source and drain. Pairing an NMOS with a PMOS transistor creates a CMOS inverter, in which the NMOS pulls the output low for a high input, and the PMOS pulls it high for a low input. The gate must retain electrostatic control of the channel to ensure clean switching. As gate lengths shrank, the drain began to compete with the gate for that control, increasing off-state leakage. Electrostatic control was restored through an architectural evolution that raised the channel into a vertical fin and wrapping the gate around three sides. Called “FinFET”, this new architecture packed more effective channel width into a smaller footprint. Further scaling made it harder to maintain both drive current and leakage within smaller cells, and reintroduced the same problems planar MOSFETs faced. Nanosheet GAAFETs close the fourth side by replacing the vertical fin with a stack of horizontal nanosheets, each surrounded by the gate. The tighter electrostatic control suppresses leakage at shorter gate lengths while stacking adds effective channel width within the cell footprint. In a FinFET process, channel width changes in discrete steps as designers must add or remove whole fins. Nanosheet width can instead be adjusted continuously within the process’s design rules. Wider sheets increase drive current, while narrower sheets reduce capacitance at the cost of drive current. Intel 18A uses stacks of four nanosheets each and varies their widths across logic and SRAM. At the process level, adding more sheets to each stack increases effective channel width and drive current, but complicates fabrication. RibbonFET vs MBCFET Samsung began GAAFET production in 2022 with SF3E, following with SF3 and now SF2. Its ‘MBCFET’ provides a useful structural comparison with Intel’s first RibbonFET implementation. [17] STEEL is digging deeper into SF2, used in the Exynos 2600, and TSMC’s GAAFET N2, used in Apple’s A20 Pro, in upcoming newsletter articles. We’re throwing some teasers on X. Let’s compare Samsung SF2’s MBCFET with Intel 18A’s RibbonFET. Subscribe Even to the untrained eye, Intel’s extra nanosheet is obvious. Intel stacks four ribbons to Samsung’s three. Samsung’s sheets are much wider in these fields, so both sheet count and width matter to the available channel perimeter. Sheet width also changes which silicon surfaces carry current. On conventional (001) silicon, wide nanosheets emphasize the broad top and bottom surfaces, favoring electron transport; the larger sidewall contribution in a narrow sheet favors hole transport. Thinner sheets improve gate control but increase confinement and scattering. This makes width and thickness part of the NMOS/PMOS balance, alongside strain and threshold voltage. [18], [19] GAAFET designs like 18A use different work-function-metal (WFM) stacks for NMOS and PMOS. Around each ribbon, a thin SiOx interfacial layer separates the silicon channel from the HfOx high-k dielectric, with La providing dipole tuning and the WFM wrapping the dielectric. NMOS uses a TiAl-based stack, while PMOS uses TiN WFM. W fills the remaining gate trench, providing a lower-resistivity path where the work-function layers are no longer needed. In this field, the PMOS stacks leave room for W between ribbons, while the NMOS stacks occupy more of those gaps. A silicon-based dielectric marks the P/N boundary, allowing the PMOS and NMOS gates, sharing the same gate trench, to be processed sequentially. Fast logic paths, retention circuits, and SRAM need a family of threshold options. Changing threshold without substantially changing device dimensions, capacitance or fabrication complexity is valuable. FinFET processes typically use different work-function-metal stacks. In a four-ribbon GAA stack, the narrow sheet-to-sheet gap limits how much WFM can fit around each channel. La in the gate dielectric creates interfacial dipoles at the SiOx/HfOx boundary, shifting effective work function and tuning threshold voltage. This gives Intel another control alongside its NMOS and PMOS WFM stacks. Low-threshold devices improve critical-path drive; higher thresholds reduce leakage elsewhere. Dipole tuning is especially useful in GAA because it changes threshold without consuming the narrow intersheet gap with thicker WFM. Precise control of La incorporation, diffusion and interface quality has long been a challenge, limiting viability in high volume production but is now seen from every leading-edge foundry. Intel’s patent describes depositing a dipole-forming oxide above HfOx and annealing it toward the interfacial oxide before completing the work-function and fill metals. This separates threshold tuning from the space available for metal. Newer research addresses the thermal cost: imec’s 2026 dipole-middle research inserts the shifter between two HfOx depositions, shortening the diffusion path while protecting SiOx during patterning. [20], [21]Matched-cut EDS, Intel 18A (left) vs. Samsung SF2 (right). Intel retains raised source/drain epi beneath its contacts, while Samsung recesses W deep into the epi to form a V-shaped Ti-lined interface. The deeper contact increases metal-to-semiconductor area and shortens the current path from the lower sheets, reducing contact and spreading resistance. It also removes epi volume and brings the contact etch closer to the channel ends. Retaining more epi preserves the material available for strain transfer, especially from SiGe into PMOS. These geometries balance contact access against stress engineering and etch margin. [22], [23] Samsung stacks three sheets to Intel’s four ribbons, and both processes use sheet width to tune drive strength. In our Samsung cross-sections, widths range roughly from 19 to 30 nm in the NPU rows and 37 to 50 nm in the CU cell. The Samsung nanosheets taper, with the widest sheet at the bottom and the narrowest at the top. Both processes use HfOx gate dielectric and Ti-based work-function stacks, with Al in the NMOS stack. In the Samsung devices shown here, the dielectric and WFM occupy the intersheet gaps, leaving W above the top sheet. Intel’s PMOS stack leaves more room between ribbons, and W fills those gaps while the thicker NMOS stack leaves W mainly in the upper trench. Gate-stack EDS maps. The W between Intel’s PMOS ribbons provides a conductive path close to the lower gates. Where WFM fills the entire gap, the gate still surrounds the channel, but voltage reaches it through the more resistive work-function films. Thinner WFM and dipole tuning preserve room for low-resistivity fill; Mo and Ru are alternative fill metals being developed for further scaling. [24] A masked, sequential WFM flow explains the different gate heights and inter-nanosheet fill. The proposed sequence below shows how separate NMOS and PMOS work-function steps produce that geometry. Enabled by the BSPDN process, Intel replaces the dense-logic silicon subfin with dielectric, removing the parasitic conduction path below the ribbons and reducing substrate-related capacitance. A retained silicon body as in classical, non-SOI, planar and FinFET designs needs junction and punchthrough-stop engineering to suppress leakage. Dielectric isolation makes that leakage less sensitive to the subfin doping profile but adds removal and fill steps. It also weakens the direct thermal path through silicon, making the contacts, metal stacks and package more important for heat extraction. [24], [26] Fluorine is concentrated around selected Intel device structures in the maps. WF6 is a standard precursor for W nucleation and fill, while barrier films protect adjacent dielectrics from fluorine attack. Low-fluorine W processes reduce the residual-F burden. Chloride-based precursors avoid introducing F during W deposition, but require control of chlorine attack, nucleation and fill quality. The integration target is a continuous, low-resistance W path with a thin protective liner and minimal chemical damage to the surrounding stack. [20], [27], [28] The SemiAnalysis STEEL teardown lab breaks down advanced datacenter and AI hardware. To learn more about our pipeline or to commission a teardown, contact sales@semianalysis.com. WE’RE HIRING: Architecture, floorplan, packaging, manufacturing, and labs experts. Opportunities from system to transistor and everywhere in between. Check out our Careers page. We measured cell height, gate pitch, metal geometry, and ribbon dimensions at the XTEM sites shown below. The tables group these dimensions by site and device polarity. Our “sheet cuts” cross the silicon channel and show the ribbons end-on. “gate cuts” run along the channel through successive gates. The 18A logic cell dimensions point to a five-track logic library while the N3E and Intel 3 cell dimensions evidence a seven-track logic library. The DDR-PHY uses wider M0 wires and much larger spacing than core logic. That trades routing density for lower wire resistance and weaker coupling between neighboring nets. The geometry suits the current delivery and coupling requirements of analog, clock, and I/O circuitry. PowerVia lets 18A combine a compact cell height with wider M0 geometry by moving the main power rails off the signal-routing tracks. That relaxes local wire scaling while preserving a small cell footprint. Cell height and gate pitch set the geometric density; pin access and routability determine how much of it a real block can use. [29] The biggest takeaway from our gate-pitch measurements is that Intel 18A compute logic and TSMC N3E GPU logic have similar density in the Bohr representative-cell model. The 18A example is 18.6% denser than the Intel 3 GPU example. Gate pitches are nearly identical across the three sites, so cell height drives most of the difference. The Bohr model combines a four-transistor NAND2 spanning three gate pitches and a 32-transistor scan flip-flop (SFF) spanning nineteen pitches, weighting their densities 60:40. The sensitivity column shows how independently changing cell height and gate pitch by ±1 nm changes the result. This compares representative cell geometries; whole-die density also depends on cell mix and placement. The 18A P-core gives M0 substantially more metal cross section than the N3E vector engine. Treating each profile as a trapezoid gives 2.63 times the area per line and 1.84 times the area after normalization by routing pitch. The larger section reduces the geometric contribution to line resistance and lowers current density for a given current. Taller and wider wires also add capacitance, so circuit delay depends on the balance of resistance and capacitance. The DDR-PHY has less metal area per routing width than the 18A core fields, while remaining above N3E. [30] Area = height × (top CD + bottom CD) / 2, including liners. Area/pitch normalizes by routing width. Taper is the symmetric sidewall angle from vertical, with the largest angle belonging to the DDR-PHY. Compute tile The measurements show how ribbon dimensions and gate-stack geometry vary across the compute tile and between NMOS and PMOS to balance channel drive, gate load and the space needed for the dielectric/WFM stack across logic, SRAM and the DDR-PHY. Width mainly changes available channel perimeter; thickness also changes electrostatic control and carrier confinement. Gate-stack thickness then determines the space left for low-resistivity fill P-core and LP E-core logic Both the P-core and LP E-core use multiple nanosheet widths. Widths are measured on high-magnification XTEMs while wider-field images demonstrate additional width choices within the LP E-core. Multiple widths are expected even within an LP E-core. Timing-critical paths, buffers and cells with different fanout need different drive strengths. The lower-magnification fields show this width diversity beyond the sites quantified in the table. L2 and L3 SRAM GAA gives SRAM designers another way to balance the pull-up (PU), pass-gate (PG), and pull-down (PD) transistors. FinFET bitcells set device strength through fin count while GAA adds nanosheet width as a sizing knob. In a 6T SRAM cell, a strong pull-down relative to the pass-gate limits read disturbance, while a strong pass-gate relative to the pull-up improves writability. During a write, the pass-gate and write driver pull the node storing “1” below the inverter trip point. During a read, the pull-down holds the node storing “0” low. Bias, threshold voltage, mismatch and assist circuitry set the remaining margin. FinFET high-current cells commonly use a PU:PG:PD fin-count pattern of 1:2:2, a device-sizing ratio rather than a current ratio. Ribbon width lets Intel balance SRAM strengths without adding whole fins. The L2 cell uses its narrowest ribbons for PU and widest for PD, improving writability and read stability respectively. Intel’s disclosed HCC operates without assist; its denser HDC uses negative-bitline write assist. Pulling the selected bitline briefly below ground increases pass-gate overdrive so it can overpower the pull-up at lower supply voltage. That buys density and low voltage writability at the cost of boosting circuitry, switching energy, and additional voltage stress that must be controlled. [31], [32] Four rectangular ribbons give the perimeter = 8 × (width + thickness), before corner rounding. PG/PU is 1.49 and PD/PG is 1.16. The L3 structures closely resemble L2 in layout and cell height. Fewer L3 nanosheet widths are tabulated because fewer high-magnification images were available. DDR PHY The DDR-PHY trades density for controlled analog behavior and reliable off-chip signaling. It contains drivers, receivers, delay circuits, and calibration logic that set drive strength, sampling time, and voltage margin. Repeated four-sheet devices with similar widths fit the use of regular transistor units for matching and programmable drive. Its wider local wiring provides room for current delivery and separation of sensitive signals, while consuming more area than a dense core-logic grid. The layout serves the memory channel’s electrical requirements as well as digital logic density. [33] The SemiAnalysis STEEL teardown lab breaks down advanced datacenter and AI hardware. To learn more about our pipeline or to commission a teardown, contact sales@semianalysis.com. WE’RE HIRING: Architecture, floorplan, packaging, manufacturing, and labs experts. Opportunities from system to transistor and everywhere in between. Check out our Careers page. Intel 3 GPU devices Vector engine logic Intel 3’s XVE logic uses two-fin PMOS and NMOS devices with power rails in M0. Its cell height and M0 pitch give a seven-track geometry, two tracks more than the 18A logic. One-fin groups also appear among the two-fin devices. Intel 3 L2 SRAM The Intel 3 L2 SRAM uses the familiar HCC sizing pattern: one PU fin, two PG fins, and two PD fins. N3E GPU devices Vector engine logic The N3E XVE field contains repeated two-fin devices with seven-track cell geometry. N3E remains a FinFET process, giving Panther Lake a direct FinFET-to-RibbonFET comparison. N3E L2 SRAM The N3E L2 SRAM uses the same PU:PG:PD fin-count pattern of 1:2:2. Panther Lake-U follows Lunar Lake’s floorplan quite closely. Both pair 4 P-cores with 4 LP E-cores and NPU, media and display engines in similar locations. Lunar Lake also uses Xe2, the direct predecessor to Panther Lake’s Xe3 GPU. This makes Lunar Lake the most direct basis for our comparisons. Arrow Lake differs in core count and uses the older Xe-LPG GPU architecture, so we only use it where it offers a more direct component-level comparison. Compute tile Panther Lake compute-tile floorplans remain sparse even months after launch. Intel 18A’s backside metal and dielectric stack must be removed without damaging the underlying structures before a clean transistor-level floorplan can be imaged. Most published die shots hide or heavily process the background, but we are quite proud of the die shot we achieved and are excited to show the work we have done. We measured the areas of the key components on the compute tile and compared them with their Lunar Lake predecessors on TSMC N3B. These help us to capture changes in block area and compare the two chips across process nodes and designs. Our total tile areas exclude the scribe-line area. The compute-plus-GPU subtotal below uses the PTL-U compute tile and GT1 GPU; it excludes the I/O tile and passive base. Individual block areas use the boundaries marked on the floorplans The compute-plus-GPU row is recomputed from the displayed PTL-U and GT1 areas. Component rows use their stated per-region counts and are not an additive partition of the whole tile. The P-core area remains almost unchanged between Lunar Lake and Panther Lake, despite L2 capacity increasing from 2.5 MiB to 3 MiB. Arrow Lake uses the same Lion Cove core as Lunar Lake but also has a 3 MiB L2. Cougar Cove fits 20% more L2 into the same P-core area. The larger private cache keeps more of each core’s working set close to its execution units, reducing access to shared L3 and DRAM. Extra capacity adds storage leakage and lookup energy, so designers balance it against avoided lower-level accesses. The shared P-core L3 cache also shrank by 14.8%. [2] Cougar Cove combines a similar footprint with Intel’s reported power-efficiency improvements. RibbonFET’s tighter channel control reduces leakage, while PowerVia reduces supply droop and allows tighter voltage guardbands. [1] Darkmont’s four-core LP E-core cluster is 5.0% smaller than Skymont’s on Lunar Lake, with most of the reduction in its L2 regions. The 1 MiB region shrank by 8.4% and the 1.5 MiB region by 14.9%. The tag arrays also use one fewer visible row. Tags identify which memory addresses the data array holds, so rearranging them changes the cache’s layout and wiring without requiring less data capacity. [2] The LP E-cores share one L2. This pools capacity and avoids duplicating all the cache machinery, but the four cores contend for its banks and bandwidth. Their separate cluster also keeps light work away from the performance cluster and its L3, allowing that larger domain to sleep. [1], [2] Cache area includes more than the storage cells. Tags identify each line, decoders select rows, sense amplifiers read the small bitline signal, and wires connect to the banks. Splitting an array into smaller sections shortens wordlines and bitlines, improving access speed, but duplicates peripheral circuits. Panther Lake’s smaller cache regions therefore reflect the complete memory implementation, including how much of each region is devoted to storage. [34] Unlike Meteor Lake and Arrow Lake, Panther Lake has no separate SoC tile. The NPU, LP E-cores, memory controllers, PHYs, media and display engines now share the compute tile. This removes an active die and keeps CPU memory traffic on one die. The cost is moving PHY and I/O-related circuitry onto 18A: drivers, receivers and analog circuits must still meet external voltage, loading and signal-integrity requirements, so their area does not shrink like dense digital logic. [1], [2] The biggest shrink comes from the NPU, which occupies 36.9% less area. NPU 5 consolidates the same total INT8 MAC count into half as many neural compute engines. Each of the three NCEs has a larger MAC array to make the complete NCE envelope 22.6% larger than an NPU 4 engine. Consolidation also halves the number of scratchpads and SHAVE DSPs, from 12 to 6. The MAC array handles matrix multiplication and convolution, while SHAVE executes vector and custom operations that fit the array poorly. [1], [2], [35] The paired floorplans identify each NCE envelope and its scratchpad, MAC, and SHAVE regions. Each measured MAC polygon is counted once per NCE in the area accounting below. The scratchpads store weights, activations, and intermediate results near the MAC arrays, allowing repeated use without fetching them again from DRAM. Halving their number delivers the largest measured area saving but leaves less local storage for the same total MAC count. Layers that no longer fit locally require smaller working tiles or more transfers of intermediate data. The benefit depends on keeping the enlarged arrays busy while managing that tighter storage budget. [36] NPU 5 also adds native FP8. Using half the operand width of FP16 reduces storage and transfer demand, helping workloads fit the smaller local memory budget. Lower precision and format-dependent range make scaling and model validation part of deployment. Hardware activation functions further reduce work that would otherwise occupy the programmable DSPs. [1], [2] Microsoft requires an NPU to deliver at least 40 TOPS for Copilot+ PCs. Both Lunar Lake and Panther Lake meet this threshold, but Panther Lake uses significantly less silicon. GPU tiles Panther Lake is Intel’s first product with Xe3, its latest GPU architecture. It offers two different GPU tiles: a smaller GT1 tile with 4 Xe3 cores on Intel 3 and a larger GT2 tile with 12 Xe3 cores on TSMC N3E. Panther Lake allows us to compare the same GPU architecture across both Intel 3 and TSMC N3E. Wildcat Lake adds a third Xe3 implementation on Intel 18A. A future newsletter will detail Xe3 and its implementation differences across all three process nodes. GT2 scales Xe3 to a different physical layout, with render slices arranged vertically instead of GT1’s horizontal arrangement. Slice placement sets the distances to shared cache banks and the D2D interface. Those wires consume area and add delay, so scaling the number of Xe cores also requires a new balance of cache placement, routing and timing. [1] What’s immediately obvious is that the GT2 tile on TSMC N3E has much smaller Xe cores than GT1. These block areas include logic, caches, and routing. An Xe core on the GT1 tile is ~69% larger than one on Lunar Lake, and ~55% larger than one on GT2. Intel 3 therefore uses substantially more area per Xe core. The block-area gap exceeds the measured logic and SRAM density gaps, bringing routing, timing targets, cell mix, and floorplan allocation into the comparison. The measured vector/matrix engine region is almost unchanged between Lunar Lake and Panther Lake’s GT2 tile. Xe3 retains eight 512-bit vector engines and eight 2048-bit XMX engines per core. Its gains also come from feeding those engines more effectively: more resident threads hide stalls, and variable register allocation lets shaders trade registers per thread against the number of threads kept active. [1] The shared L1/SLM capacity increased by 33% from 192 KiB to 256 KiB, while its area increased only 5%, raising effective density by 27%. L1 retains reused cache lines, while software-managed SLM lets a thread group share data locally. Both reduce traffic to more distant memory. Allocating more SLM per group can also limit how many groups reside on a core at once. [1], [37] The GT1 tile carries 4 MiB of L2 against 16 MiB on the GT2 tile. GT1 divides its L2 cache into four 1 MiB banks, while GT2 uses eight 2 MiB banks. Each bank contains 128 macros, but each N3E macro stores 16 KiB, twice the Intel 3 macro’s 8 KiB capacity. The N3E macro is only 54% larger while holding twice as many bits, giving it 30% higher density: ~23.7 Mbit/mm² versus 18.3 Mbit/mm². Including bank-level circuitry, the gap widens to ~16.9 Mbit/mm² on GT2 versus ~10.4 Mbit/mm² on GT1. GT2 gains density with its macros storing more bits per unit area, and those macros occupy more of each cache bank. Larger macros spread decoder and sense-amplifier overhead across more storage, while a more compact bank layout reduces the share spent on control and routing. The compromise is longer wordlines and bitlines that carry more capacitance. [34] I/O tile Panther Lake uses two I/O tile variants, both fabricated on TSMC N6. The smaller one provides 4 PCIe 5.0 and 8 PCIe 4.0 lanes and serves lower-tier systems as well as those without a discrete GPU, while the larger one adds 8 PCIe 5.0 lanes, bringing the total to 20 lanes, for discrete-GPU connectivity. Panther Lake SKUs with the larger 10- or 12-Xe GPUs use the smaller I/O tile. [38] The smaller I/O tile adds a PCIe 4.0 block and a Thunderbolt block to Lunar Lake’s I/O layout, providing four additional PCIe 4.0 lanes and another Thunderbolt 4 port. Its repeated N6 blocks retain nearly identical areas and layouts. Reusing these proven PHYs and controllers avoids porting and requalifying external interfaces on 18A, where faster digital logic offers less benefit to circuits constrained by the off-chip link. [38] SemiAnalysis’s teardown lab (STEEL) dives deep into the world’s advanced datacenter and AI hardware. To learn more about our pipeline or to commission a teardown, contact sales@semianalysis.com. We’re hiring technical experts from system to transistor and everywhere in between. Check out our Careers page. Panther Lake offers scalability and modularity through its disaggregated packaging that partition compute, GPU, and I/O silicon into separate tiles allowing for a suite of tile configurations. This partitioning makes the package part of Intel’s node economics as it determines how much leading-edge wafer area each product consumes, which functions can remain on other processes, and how much configuration freedom Intel can offer from a shared set of tiles. Furthermore, fabricating the compute and GPU tiles separately confines the new 18A process to the compute tile and allows graphics and I/O to use other, more established, and more cost-effective processes. For Panther Lake, the GPU and I/O tiles are assembled alongside the compute tile on a passive silicon base using Foveros-S. Intel’s current technology brief lists a nominal 36 µm pitch for Foveros-S. Through-silicon vias (TSVs) in the base connect the fine wiring above to the larger package connections below. The functional tiles sit side by side on that passive base in a 2.5D configuration. [39] Our cross-section through the compute and GPU tiles shows the package’s wiring hierarchy. Microbumps connect each active tile to the passive silicon base; its fine redistribution layer (RDL) carries the short, dense tile-to-tile links. TSVs carry connections through the base to the package substrate, which fans them out to the much coarser motherboard solder joints. The base supplies interconnect, while computation remains in the active tiles above it. [39] At the compute-tile edge, the higher-magnification inset shows a local microbump spacing of approximately 25.24 µm and a feature width of 12.33 µm. These local spacings are finer than Intel’s nominal Foveros-S value. The X-ray fields further confirm tighter neighboring bumps, consistent across every die-to-die area found on each tile. Additional X-ray analysis is offered after the paywall. Putting the memory controller beside the CPU removes the D2D transfer that CPU memory requests required in Meteor Lake and Arrow Lake. This avoids the extra transmitter, receiver, and link traversal, saving interface energy and latency. Panther Lake’s separate GPU still crosses a D2D link to reach DRAM, so its larger local caches also help contain package traffic. [1], [40] Smaller dies are less likely to contain a random fatal defect, and screening them before assembly prevents one bad tile from consuming a complete package of good silicon. Reuse also spreads design and qualification work across more products. Against those gains, Intel pays for the passive base, D2D circuits, extra bonding and test steps, and losses during assembly. Cost per working product across the portfolio captures the combined effect of wafer yield, reuse, test, and assembly. [29] Wildcat Lake packaging Intel launched Core Series 3, formerly Wildcat Lake, on 16 April 2026 for value mobile and edge systems. Wildcat Lake keeps 18A but removes the passive base and combines more functions on one die to simplify the package. The two products therefore reveal two distinct ways to commercialize the same leading-edge process. [41] Wildcat Lake’s 18A die combines up to two Cougar Cove P-cores, four Darkmont LP E-cores, two Xe3 cores and a smaller NPU. A separate platform-controller die supplies I/O, connected through UCIe, Intel’s first processor implementation of the standard. Consolidating graphics remove a tile boundary and the passive base, reducing assembly complexity for a modest-bandwidth value product. It also ties CPU and graphics scaling to the same die, giving up Panther Lake’s ability to swap in a much larger GPU. [42], [43] In July 2021, Intel CEO Pat Gelsinger set out an ambitious process roadmap aimed at regaining performance leadership by 2025, later described as five nodes in four years. Five years and one CEO later, Intel’s comeback story is not as unambiguously positive as Pat may have hoped. [44], [45] Intel once set the pace for process technology, bringing high-k metal gate technology and FinFETs into volume production years ahead of the rest of the industry. Its 22 nm FinFET process reached consumers with Ivy Bridge in 2012. [46] Intel’s integrated device manufacturing (IDM) model allowed its architects and process engineers to co-optimize products and processes. Starting with Sandy Bridge, Intel dominated x86, while AMD struggled with Bulldozer. That lead faltered at 14 nm and broke at 10 nm. Intel targeted a massive 2.7× density increase, but the node arrived years late and required several revisions before it could support Intel’s full lineup. This delay forced Intel to stretch 14 nm across six generations, while TSMC moved ahead in process technology and AMD recovered in x86. By 2019, Intel was still shipping 14 nm across most of its product stack, with its 10 nm client ramp focused on Ice Lake mobile processors. Meanwhile, TSMC was shipping N7 and N7+, and AMD’s Zen 2 compute chiplets used N7 to raise core counts and improve efficiency. Intel’s process failures were central to its decline, but unsound business decisions furthered their downward slide. Product delays compounded product mistakes, pushing client, server, and FPGA roadmaps off schedule. Several attempts to enter AI (Nervana and Gaudi) and networking (Tofino) also failed to establish lasting businesses. Intel’s recovery has focused on consumer CPUs and advanced packaging. Tiger Lake, Alder Lake, Lunar Lake and now Panther Lake have restored Intel’s consumer roadmap. On the process side, Intel 4 shipped with Meteor Lake, Intel 3 with Granite Rapids and Sierra Forest, and Intel 18A with Panther Lake. Intel has also made advanced packaging part of its foundry offering. However, Intel is still playing catch-up in servers. Several Xeon generations arrived years late and trailed contemporary AMD and Arm server CPUs in performance, efficiency, and core count. The process roadmap is back, but Intel does not hold the same process-technology leadership position it held prior to 10 nm. The introduction of gate-all-around nanosheets and backside power delivery are two of the biggest changes to transistor integration in a decade. Intel took on both changes at once: 18A paired its first RibbonFET with PowerVia in Panther Lake. Panther Lake is a substantial manufacturing milestone. Our cross-sections show how RibbonFET and PowerVia reshape local contacts and wiring, while the floorplans show where architectural consolidation and process choices save area. A sustained competitive lead depends on product performance, cost, yield, and the next implementation. The SemiAnalysis STEEL teardown lab breaks down advanced datacenter and AI hardware. To learn more about our pipeline or to commission a teardown, contact sales@semianalysis.com. WE’RE HIRING: Architecture, floorplan, packaging, manufacturing, and labs experts. Opportunities from system to transistor and everywhere in between. Check out our Careers page.. [1] Intel, “Intel Tech Tour 2025 Panther Lake recap,” Doc. 866361, 2025. Accessed: Sep. 14, 2026. [Online]. Available: https://cdrdv2-public.intel.com/866361/ITT_2025_Panther_Lake_Recap1.pdf [2] Intel, “Core Ultra processors Series 3 datasheet, volume 1,” Doc. 872188-001, Jan. 2026. Accessed: Sep. 14, 2026. [Online]. Available: https://cdrdv2-public.intel.com/872188/872188-001.pdf [3] N. Horiguchi and E. Beyne, “Backside power delivery: How to power chips from the backside,” imec Reading Room, Nov. 25, 2022. Accessed: Sep. 14, 2026. [Online]. Available: https://www.imec-int.com/en/articles/how-power-chips-backside [4] W. Hafez et al., “Intel PowerVia technology: Backside power delivery for high density and high-performance computing,” in Proc. IEEE Symp. VLSI Technol. Circuits, Kyoto, Japan, Jun. 2023, pp. 1–2, doi: 10.23919/vlsitechnologyandcir57934.2023.10185208. [5] K. Fischer et al., “Intel 18A platform technology featuring RibbonFET (GAA) and PowerVia for advanced high-performance computing,” in Proc. Symp. VLSI Technol. Circuits, Kyoto, Japan, Jun. 2025, pp. 1–3, doi: 10.23919/vlsitechnologyandcir65189.2025.11075006. [6] K.-F. Cheng, C.-L. Teng, H.-Y. Huang, and H.-C. Chen, “Metal oxide composite as etch stop layer,” U.S. Patent 12 176 247 B2, Dec. 24, 2024. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.google.com/patent/US12176247B2/en [7] S. W. King, “Dielectric barrier, etch stop, and metal capping materials for state of the art and beyond metal interconnects,” ECS J. Solid State Sci. Technol., vol. 4, no. 1, pp. N3029–N3047, 2015, doi: 10.1149/2.0051501jss. [8] A. V. Mule’, D. J. Towner, D. Seghete, C. R. Ryder, and A. A. Gonzalez, “Multi-layer etch stop layers for advanced integrated circuit structure fabrication,” U.S. Patent Appl. 20220102343 A1, Mar. 31, 2022. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.google.com/patent/US20220102343A1/en [9] C.-C. Wang and J. H. Wang, “Interconnect structure with dielectric cap layer and etch stop layer stack,” U.S. Patent Appl. 20230253247 A1, Aug. 10, 2023. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.google.com/patent/US20230253247A1/en [10] Y. J. Wu, K.-F. Cheng, C.-C. Lee, H.-K. Chang, and H.-Y. Huang, “Etch stop layer for interconnect structures,” U.S. Patent Appl. 20240332070 A1, Oct. 3, 2024. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.google.com/patent/US20240332070A1/en [11] M. G. Rainville, N. Shankar, K. S. Reddy, and D. M. Hausmann, “Deposition of aluminum oxide etch stop layers,” U.S. Patent Appl. 20180197770 A1, Jul. 12, 2018. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.justia.com/patent/20180197770 [12] J. S. Leib et al., “Conductive lines having molybdenum liner and tungsten fill for advanced integrated circuit structure fabrication,” U.S. Patent Appl. 20240429126 A1, Dec. 26, 2024. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.justia.com/patent/20240429126 [13] M. Naik, “Cobalt enables power and performance scaling at single-digit logic nodes,” Applied Materials, Dec. 17, 2018. Accessed: Sep. 14, 2026. [Online]. Available: https://www.appliedmaterials.com/us/en/blog/blog-posts/cobalt-enables-power-and-performance-scaling-at-single-digit-logic-nodes.html [14] Lam Research, “Breaking through AI’s invisible barrier with molybdenum,” Lam Research Newsroom, May 6, 2025. Accessed: Sep. 14, 2026. [Online]. Available: https://newsroom.lamresearch.com/molybdenum-metallization-ai-revolution [15] P. Yashar, G. Malyavanatham, and H. Vijwani, “Integrated circuit interconnect structures with niobium barrier materials,” U.S. Patent Appl. 20240112951 A1, Apr. 4, 2024. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.google.com/patent/US20240112951A1/en [16] Applied Materials, “Applied Materials unveils chip wiring innovations for more energy-efficient computing,” Applied Materials Investor Relations, Jul. 8, 2024. Accessed: Sep. 14, 2026. [Online]. Available: https://ir.appliedmaterials.com/news-releases/news-release-details/applied-materials-unveils-chip-wiring-innovations-more-energy [17] Samsung, “Samsung showcases AI-era vision and latest foundry technologies at SFF 2024,” Samsung Semiconductor, Jun. 13, 2024. Accessed: Sep. 14, 2026. [Online]. Available: https://semiconductor.samsung.com/news-events/news/samsung-showcases-ai-era-vision-and-latest-foundry-technologies-at-sff-2024/ [18] C. W. Yeung et al., “Channel geometry impact and narrow sheet effect of stacked nanosheet,” in Proc. IEEE Int. Electron Devices Meeting (IEDM), Dec. 2018, pp. 28.6.1–28.6.4, doi: 10.1109/iedm.2018.8614608. [19] S. Mochizuki et al., “Evaluation of (110) versus (001) channel orientation for improved nFET/pFET device performance trade-off in gate-all-around nanosheet technology,” in Proc. Int. Electron Devices Meeting (IEDM), Dec. 2023, pp. 1–4, doi: 10.1109/iedm45741.2023.10413854. [20] D. G. Ouellette et al., “Fabrication of gate-all-around integrated circuit structures having molybdenum nitride metal gates and gate dielectrics with a dipole layer,” U.S. Patent 12 051 698 B2, Jul. 30, 2024. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.google.com/patent/US12051698B2/en [21] imec, “Advancing the CFET-based device roadmap with novel integration modules and standard-cell architectures,” imec, 2026. Accessed: Sep. 14, 2026. [Online]. Available: https://www.imec-int.com/en/articles/advancing-cfet-based-device-roadmap-novel-integration-modules-and-standard-cell [22] A. Reznicek, X. Miao, C. Lee, and J. Zhang, “Wrap around contact for nanosheet source drain epitaxy,” U.S. Patent 11 302 813 B2, Apr. 12, 2022. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.google.com/patent/US11302813B2/en [23] J. A. Kittl, J. G. Hong, D. R. Palle, and M. S. Rodder, “Methods for varied strain on nano-scale field effect transistor devices,” U.S. Patent 9 978 833 B2, May 22, 2018. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.google.com/patent/US9978833B2/en [24] S. Gandikota et al., “Method of reducing metal gate resistance for next generation NMOS device application,” Int. Patent Appl. WO 2024/137272 A1, Jun. 27, 2024. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.google.com/patent/WO2024137272A1/en [25] J. Zhang et al., “Full bottom dielectric isolation to enable stacked nanosheet transistor for low power and high performance applications,” in Proc. IEEE Int. Electron Devices Meeting (IEDM), Dec. 2019, pp. 11.6.1–11.6.4, doi: 10.1109/iedm19573.2019.8993490. [26] C. Yoo, J. Chang, Y. Seon, H. Kim, and J. Jeon, “Analysis of self-heating effects in multi-nanosheet FET considering bottom isolation and package options,” IEEE Trans. Electron Devices, vol. 69, no. 3, pp. 1524–1531, Mar. 2022, doi: 10.1109/ted.2022.3141327. [27] S. S. Pradhan, D. B. Bergstrom, J.-S. Chun, and J. Chiu, “Tungsten gates for non-planar transistors,” European Patent Appl. EP 3 506 367 A1, Jul. 3, 2019. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.google.com/patent/EP3506367A1/en [28] L. Schloss and X. Ba, “Tungsten films having low fluorine content,” U.S. Patent 9 754 824 B2, Sep. 5, 2017. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.google.com/patent/US9754824B2/en [29] Intel Foundry, “Accelerating AI and HPC with advanced process and packaging technologies,” Doc. 367370-001US, Jun. 2026. Accessed: Sep. 14, 2026. [Online]. Available: https://www.intel.com/content/dam/www/central-libraries/us/en/documents/2025-11/intel-foundry-hpc-ai-brief.pdf [30] M. Lofrano, X. Chang, H. Oprins, and Z. Tokei, “Mitigating the thermal bottleneck in advanced interconnects,” imec, Sep. 28, 2023. Accessed: Sep. 14, 2026. [Online]. Available: https://www.imec-int.com/en/articles/mitigating-thermal-bottleneck-advanced-interconnects [31] S. Bair, “Intel at ISSCC 2025: Navid Shahriari invited talk, eight papers, forums, panelist & product details,” Intel Community, Feb. 19, 2025. Accessed: Sep. 14, 2026. [Online]. Available: https://community.intel.com/t5/Blogs/Tech-Innovation/Edge-5G/Intel-at-ISSCC-2025-Navid-Shahriari-Invited-Talk-Eight-Papers/post/1667592 [32] D. Chandra, E. Potladhurthi, D. R. S. Reddy, and K. S. Rengarajan, “Tunable negative bitline write assist and boost attenuation circuit,” U.S. Patent Appl. 20160203857 A1, Jul. 14, 2016. Accessed: Sep. 14, 2026. [Online]. Available: https://patents.google.com/patent/US20160203857A1/en [33] Synopsys, “Advantages of firmware-based training in high-speed DDR IP,” Synopsys IP Technical Bulletin. Accessed: Sep. 14, 2026. [Online]. Available: https://www.synopsys.com/articles/firmware-based-training-ddr-ip.html [34] N. Muralimanohar, R. Balasubramonian, and N. P. Jouppi, “CACTI 6.0: A tool to understand large caches,” Hewlett-Packard Laboratories, 2009. Accessed: Sep. 14, 2026. [Online]. Available: https://users.cs.utah.edu/~rajeev/cacti6/cacti6-tr.pdf [35] Intel, “Intel Tech Tour 2024 Lunar Lake AI hardware accelerators,” Doc. 824436, 2024. Accessed: Sep. 14, 2026. [Online]. Available: https://cdrdv2-public.intel.com/824436/2024_Intel_Tech%20Tour%20TW_Lunar%20Lake%20AI%20Hardware%20Accelerators.pdf [36] Y.-H. Chen, J. Emer, and V. Sze, “Eyeriss: A spatial architecture for energy-efficient dataflow for convolutional neural networks,” in Proc. 43rd ACM/IEEE Annu. Int. Symp. Comput. Archit. (ISCA), Jun. 2016, pp. 367–379, doi: 10.1109/isca.2016.40. [37] Intel, “Introduction to the Xe-HPG architecture,” Doc. 758306, Nov. 4, 2022. Accessed: Sep. 14, 2026. [Online]. Available: https://www.intel.com/content/www/us/en/developer/articles/technical/introduction-to-the-xe-hpg-architecture.html [38] Intel, “Core Ultra processors Series 3 for the edge,” Doc. 855291. Accessed: Sep. 14, 2026. [Online]. Available: https://cdrdv2-public.intel.com/855291/Intel%C2%AE%20Core%E2%84%A2%20Ultra%20Processors%20Series%203%20for%20Edge%20Overview_2.pdf [39] Intel Foundry, “Foveros technology brief,” Doc. 366411-001US, Feb. 2026. Accessed: Sep. 14, 2026. [Online]. Available: https://www.intel.com/content/dam/www/central-libraries/us/en/documents/2025-07/foveros-25d-product-brief.pdf [40] D. Das Sharma, “UCIe: Building an open chiplet ecosystem,” UCIe Consortium, 2022. Accessed: Sep. 14, 2026. [Online]. Available: https://www.uciexpress.org/_files/ugd/0c1418_c5970a68ab214ffc97fab16d11581449.pdf [41] Intel, “Intel launches Intel Core Series 3 processors,” Intel Newsroom, Apr. 16, 2026. Accessed: Sep. 14, 2026. [Online]. Available: https://www.intel.com/content/www/us/en/newsroom/news/client-computing/intel-launches-intel-core-series-3-processors-changing-the-game-for-everyday-computing.html [42] Intel, “Core Series 3 launch press deck,” Apr. 2026. Accessed: Sep. 14, 2026. [Online]. Available: https://download.intel.com/newsroom/2026/Intel-Core-Series-3/Intel-Core-Series-3-Launch-Press-Deck.pdf [43] Intel, “Intel outlines architectures for agentic AI at Hot Chips 2026,” Intel Newsroom, Aug. 24, 2026. Accessed: Sep. 14, 2026. [Online]. Available: https://www.intel.com/content/www/us/en/newsroom/news/client-computing/intel-outlines-architectures-for-agentic-ai-at-hot-chips-2026.html [44] Intel, “Intel accelerates process and packaging innovations,” Intel Investor Relations, Jul. 26, 2021. Accessed: Sep. 14, 2026. [Online]. Available: https://www.intc.com/filings-reports/all-sec-filings/content/0001193125-21-224438/d199788dex991.htm [45] Intel, “Intel reports third-quarter 2021 financial results,” Intel Investor Relations, Oct. 21, 2021. Accessed: Sep. 14, 2026. [Online]. Available: https://www.intc.com/news-events/press-releases/detail/1505/intel-reports-third-quarter-2021-financial-results [46] Intel, “3rd generation Intel Core processors bring exciting new experiences and fun to the PC,” Intel Investor Relations, Apr. 23, 2012. Accessed: Sep. 14, 2026. [Online]. Available: https://www.intc.com/news-events/press-releases/detail/577/3rd-generation-intel-core-processors-bring-exciting
Hippo's AI-Powered Claims Model Delivers Strong Customer Satisfaction Results as its Homeowners Business Grows
/PRNewswire/ -- As Hippo Holdings Inc. (NYSE: HIPO) expands its homeowners business nationally, it is scaling its claims operations through a combination of AI...
AMD Ryzen 7 9800X3D Memory Performance Scaling
The AMD Ryzen 7 9800X3D set new benchmarks in gaming performance with its impressive 96 MB L3 V-Cache. But how does memory latency factor into this powerhouse's performance? Our latest analysis dives deep into memory performance scaling, and could change how gamers approach system optimization.
Even the RTX 5090 can't hit 60FPS in Control Resonant without upscaling
TechPowerUp testing shows Control Resonant can drop the RTX 5090 to 48 FPS at native 4K with ray tracing, making upscaling essential.
Engrams Embedding Entendre: Codesign for Efficient DRAM/SSD Offloading
Engram extends standard token embeddings with learned multi-token lookups. Recurring local patterns retrieve vectors directly, reducing the need to reconstruct them through attention and feed-forward layers. With Engram model architecture optimization, it allows for lower HBM capacity to be needed for models at the same quality. This does not mean there won’t be an insane demand for HBM but it just means that model architecture will continue to innovate around constraints. This model architecture design is naturally codesigned for parameter offloading: each token accesses a few embedding rows whose addresses depend on token IDs, not hidden states. The runtime can prefetch those rows from host DRAM while earlier layers compute, keeping the table outside HBM without transferring entire weight matrices. Our Memory model contains our latest estimates of quarter by quarter HBM, DRAM, & NAND supply and demand. Offloading frees HBM for model weights and KV cache, potentially supporting larger batches or more concurrent sessions. When DRAM becomes the next constraint, NVMe offers another tier. Recommendation systems already cache frequently or recently accessed embedding rows in faster memory while backing colder rows with SSDs. After NVIDIA roadmap had to change due to massively despec’ing Rubin Ultra from 1024GB to now ~200GB of HBM per chip, model architecture optimizations like emgram maybe helpful. Our DeepSeek-V4.1-Flash configuration uses roughly 189 GiB of memory for Engram. We replace it with a memory-mapped (mmap) file and measure serving performance with offload to SSD. Later on in our report, we will show our Engram offloading experiments along with the official InferenceX agentic inference serving results on Engram models like DeepSeekv4.1 Flash across all 6 NVIDIA GPU SKUs along with MI355X. Unsurprisingly, the CUDA Moat is still mogging MI355X on the ultra popular DeepSeekV4.1 Flash model. We also show how even on high capacity HBM SKUs, offloading emgrams to DRAM could result in even better performance for most of the pareto than keeping the emgram in HBM. Our benchmark has been widely reproduced, validated and/or supported by almost every major buyer of compute from Google Cloud to Microsoft Azure to Oracle, to Meta and many more. Furthermore, it has the support of the ML community including from vLLM, LMCache, SGLang, PyTorch, Huggingface and the support of major labs like OpenAI, MiniMax, ZAI, Qwen, Moonshot Kimi, etc. Star the InferenceX GitHub repository if you find the open-source benchmark and data useful!. InferenceX is the only inference benchmark in the world to have TPUv7, Jalapeño, Nvidia Rubin NVL72, AMD, and soon, SambaNova and Trainium. Due to how realistic AgentX scenario is to real world agentic inference workloads, AMD has committed to collaborating on MI455X UALoE72 too. DeepSeek did not release the original paper’s two trained Engram models. We replicated its setup on fineweb-edu using the released code and training hyperparameters, at an estimated 6E18 FLOPs per run. We observed the same U-shape scaling: Engram improved performance over pure MoE baselines. We also reproduced DeepSeek’s results, where earlier-layer representations with Engram resembled those of later layers. Like the original Engram paper, it is possible to probe Engram’s gate scores to see what n-grams DeepSeek-V4.1-Flash makes the most use of. Our gate scan finds names, code fragments, relational phrasing, and boilerplate. These examples prioritize interesting-ness over gate strength. One unexpected result was Wright : Ace Attorney. These examples suggest learned memory optimizes the training objective, not a judgment of which facts deserve storage. Licenses, bibliography fragments, API scaffolding, and website furniture can provide prediction shortcuts, so the value of additional Engram capacity may depend on what survives data preparation. This does not show that table capacity is “wasted”: the evaluation-corpus scan establishes neither training exposure nor the capacity occupied by each category. For offloading, strong gates do not identify cache-hot rows. Low gates do not automatically save reads either: computing the gate requires the retrieved key and undoes the performance gain of a fused kernel. Skipping reads would require a separate usefulness predictor before retrieval. In the original paper’s inference-time ablation, factual-knowledge benchmarks retained just 29–44% of their original performance, while reading comprehension retained 81–93%. This is due to the training–inference mismatch. The resulting degradation therefore measures this trained model’s dependence on Engram, not the performance difference between models trained with and without it. In our ablations, suppressing Engram worsens token likelihood across all evaluated domains, especially encyclopedia text and several code corpora. Surprisingly, GSM8K accuracy stays within measured run-to-run variation and removing Engram has no effect. Engram is not a detachable dictionary beside an unchanged MoE. Removing it changes downstream features and expert selection. We tested whether rerouting hurts or compensates by holding tokens fixed in a teacher-forced experiment on CRUXEval, a code-reasoning benchmark of small Python functions where the model predicts a function’s output from its code and an input, and scoring the reference answer. Removing Engram raised answer loss from 0.2848 to 0.3093 bits/token. Forcing the ablated model to use the original Engram-on expert choices made it worse still, at 0.3375 bits/token. Rerouting partially compensates for the missing memory. Memory features and expert selection work together, rather than following a clean “memory stores facts; experts reason” division. On the same CRUXeval, removing Engram during either phase reduced accuracy and increased generated tokens; removing it throughout produced the largest changes. Keeping Engram for prefill leads to more correct answers than keeping for decode likely due to semantically richer KV cache transferred to decode workers, allowing it to mitigate some of the performance loss. Engram’s table is large, but each lookup is small. DeepSeek-V4.1-Flash requests 24 rows at each of two Engram layers, about 12.4 KiB per processed token position across the model, or 3.1 KiB per GPU when split across four GPUs. Currently as of Day 7 since Model Release, MI355X is still 2-4x worse performance per dollar compared to B200 even when normalized by Mi355X’s lower TCO. Our full total cost of ownership breakdown comes from our AI Cloud TCO Model along with monthly market surveys of over 100+ gpu clouds & gpu cloud customers. On the Day 0 release of DeepSeekv4.1 Flash, NVIDIA vLLM works out of the box with zero issues across all 6 SKUs: H100, H200, B200, B300, GB200, GB300! This was thanks to the amazing work by the NVIDIA & Interact teams! In comparison, AMD vLLM did not work on day 0 for DeepSeekv4.1 Flash.Source: SemiAnalysis InferenceX AMD’s vLLM documentation points to using vllm/vllm-openai-rocm:deepseekv41-flash-0909, but from hour 0 of the model release to hour 23, AMD has not publicly released the image. AMD claims, “SPEED IS THE MOAT,” yet it has still not released it by the 23rd hour. We wish that, going forward, the AMD team has a better process for hour 0 model releases. Eventually when they did publicly release for “day 0” image support, performance-wise, it is currently up to 14.8x worse perf per dollar than H200 and up to 42x worse perf per dollar than B200/B300. The power of the CUDA MOAT is NVIDIA’s collaboration with its massive 6 million-developer community ecosystem including most of the vLLM & SGLang & Tokenspeed maintainers which means that CUDA is optimized on day 0. Overall, AMD did make significant improvements but the performance per dollar is still currently 2-4x worse than B200. AgentX Engram DRAM Offloading Improving Performance The HBM and DRAM offload use the same GPU kernel to select and dequantize rows. With HBM, it reads GPU memory; with Unified Virtual Addressing (UVA), it reads pinned host memory directly. Both support full decode graphs. Moving the table into HBM accelerates only the sparse lookup, leaving decoder computation and communication unchanged, resulting in little overall benefit while consuming memory otherwise available to KV cache. Another benefit of Engram offloaded to DRAM which means you can reduce the communication overhead by using less HBM GPUs per replica. For example, when enabling Engram offloading on B300, we are able to switch from TP4 to now TP2 which improves the pareto curve by up to 1.6x. When iso-model quality, less HBM is required as the engrams could be offloaded to host DRAM. Thus HBM bandwidth matters way more than HBM capacity. For inference workloads where memory bandwidth matters the most, 4-hi HBM provides the best $/bandwidth and therefore lowest cost per token. If China continues to make more and more revolutionary model architecture innovations, soon it could potentially 0Hi HBM stacks. [ ](https://newsletter.semianalysis.com/p/long-live-the-short-king-why-4-hi) [ Long Live the Short King: Why 4-hi HBM Wins ](https://newsletter.semianalysis.com/p/long-live-the-short-king-why-4-hi) Myron Xie, Bryan Shan, and 3 others · Sep 13 [ Read full story ](https://newsletter.semianalysis.com/p/long-live-the-short-king-why-4-hi) Moreover, on B300 and week-0 stack, moving the Engram table back to HBM did not improve results and stay within run-to-run variance. This is a result of the work optimizing DRAM offload, such as async, and overlap. SSD Offloading On B200, we created a unoptimized vLLM fork and stored the Engram tables in memory-mapped files on local SSDs. File backing lets the OS reclaim table pages when other applications need RAM. Pages already cached in memory can be served without reading the SSD again. Furthermore, note that we were unable to turn on GDS The unoptimized SSD implementation changes how rows reach the GPU. It copies row IDs to the CPU, deduplicates them, gathers the requested rows into pinned buffers, copies those rows back to the GPU and dequantizes them. This work runs between segments of the GPU execution graph. Native UVA performs row selection and dequantization directly on the GPU, avoiding the CPU round trip. A warm filesystem cache removes physical SSD reads, but leaves the coordination, row gathering and transfers. This is why a file already cached in RAM can still perform worse than a pinned DRAM table. The comparison measures the whole serving path; it does not separate the time spent on each of these operations. B200 DRAM dominates both measured SSD serving curves in total tokens per dollar and P90 interactivity. Near 125 tokens/s/user, DRAM delivers 121 million total tokens per dollar versus 52 million for SSD. For production serving, SSD offloading is likely not worth the tradeoff. On the B200 configurations we measured, SSD offloading loses on both measures: every observed SSD point has a DRAM alternative that delivers higher P90 interactivity and more total tokens per dollar. The only points where Cheaper storage does not automatically produce a cheaper inference service. Moving Engram to SSD leaves the same four expensive GPUs and the rest of the server in place. Reclaiming RAM only creates an economic benefit if it enables a cheaper server configuration or additional useful capacity. The current unoptimized path provides neither benefit, and the filesystem cache still consumes RAM when table pages are resident. Next we will look at the mechanisms and specific implementation of ngrams from DeepSeekv4.1 Flash, LongCat, Qwen3.8 Flash Next.
Gears Of War: E-Day Defies Unreal Engine 5 Expectations, As 4K@60FPS Needs An RTX 5070 Ti Instead Of Flagship GPU
Gears of War: E-Day PC system requirements reveal tame Ultra specs for 4K 60 FPS with upscaling on Unreal Engine 5
Ukrainian defence tech seeks new ways to scale up
Since Russia's full-scale invasion in February 2022, Ukraine's defence sector has expanded several-fold, becoming a significant source of military innovation, especially in unmanned systems, counter-drone technologies, electronic warfare and AI-driven combat systems. But finding the funds and personnel to scale up and to reach international markets remains a challenge.Spurred on by the demands of the front-line, Ukrainian defence companies say they can develop a minor component in a couple of weeks, or major hardware or software solutions within a year. This is several times faster than the usual development cycles in the sector. Tencore, founded in 2024, is now one of Ukraine’s largest providers of unmanned ground vehicles, well exceeding the production of NATO members in this category. Odd Systems, a Ukrainian combat cameras developer founded in 2023, launched its first thermal camera in 2024. Less than two years later, it is developing the fourth generation of the product, one that has evolved into a full vision system.According to Michael Singer, managing partner at Verne Capital, a venture capital fund supporting defence companies in Europe, this speed is down to the short feedback loop between the developer and the end user. “Ukrainian companies develop technologies in close interaction with soldiers and military units, test them under real combat conditions and rapidly incorporate feedback into the next generation of the product,” he says. “Sometimes we need to react to feedback and deploy an upgrade within hours,” says Polina Tymchenko, the spokesperson for Himera, which develops electronic warfare-resistant walkie-talkies. “It means that, as a company, you have to have a higher tolerance for risk, and also design your products in a way that fosters those quick changes.”Getting a product deployed by a combat command involves no paperwork other than a purchase order, says Perry Boyle, co-founder and chief executive of Mits Capital, a Ukrainian-US investment group focused on attracting capital for Ukraine's defence tech sector. “New weapons can be developed and deployed based largely on handshakes.”Nevertheless, the speed of development is largely subject to the level of funding, says Iryna Sukhomlyn, head of business development at RMachine, a developer of unmanned ground systems. “Financial support needs to start much earlier, particularly at technology readiness level 3-5, not only when most of the technological risk has already been removed,” she says.In 2025, Ukrainian defence tech companies attracted about $129 million in investment, with the true figure likely higher given the share of undisclosed security-sensitive deals, according to a report published by the Kyiv School of Economics. These investments were primarily concentrated in the AI and software segment, as they are considered a lower financial risk.Scaling production is a major barrierAccording to Iryna Supruniuk, business development director at Odd Systems, proving a defence product in Ukraine is not the end of the development process, but the beginning of a continuous evolution, since operational requirements change rapidly. “Continuous iteration remains our highest priority,” she says.It's also when production must scale up, which represents a major hurdle. “Scaling production requires capital, manufacturing capacity, components, certification, procurement access and reliable supply chains,” Singer says. Scale-up is also hampered by demographic pressures, especially a shortage of young specialists needed to expand engineering teams. Companies have reported struggling to replace specialists lost due to Russian attacks, mobilisation and emigration. In order to meet this scale-up challenge and finance arms production, begun allowing defence technologies to be exported, assuming it has an excess capacity. It has also announced the creation of ten defence and weapons export centres across Europe by the end of 2026.According to Singer, a technology that has demonstrated its value under real combat conditions has a fundamentally different credibility than a product that has only been tested in a laboratory or training environment. But combat validation alone doesn’t automatically create an international business. “Companies still need access to foreign defence ministries and procurement organisations, international certification, financing for production scale-up and, in many cases, local industrial partners,” he says.Already in 2026, the EU has signed several agreements with Ukraine that aim at joint defence technology development and production. These include allowing Ukraine to join the European Defence Fund, signing a financing agreement under the European Defence Industry Programme, and setting up the EU-Ukraine Drone Alliance.For Singer, it is important that European defence companies don’t just see their Ukrainian partners as technology suppliers, but rather as potential innovation partners. “Ukrainian companies bring speed, operational experience and an exceptional ability to iterate. European partners can provide industrial capacity, capital, established supply chains and access to European procurement markets.”Procurement possibilitiesWhen it comes to exports, the procurement and compliance processes involved require time, taking months or even years, and money. In other words, resources that small defence tech companies need if they are to improve and scale up their technologies. According to Sukhomlyn of RMachine, if the priority is to rapidly increase available defence capabilities, then the processes need to reflect that urgency. “If Europe wants defence innovation at wartime speed, its administrative processes also need to operate much closer to wartime speed,” she says, highlighting a need for support with certification, market access and mutual recognition of relevant testing. “Otherwise, we risk having excellent technology available but spending years making it administratively eligible to be purchased.”Related articlesRussia widens targeting of European research centresEU must support research managers in central and eastern Europe, says Visegrad fund headThis is a lesson Europe’s defence sector would do well to learn, Singer says. “Defence innovation needs to be organised around the speed and requirements of the end user, not around the speed of the bureaucracy.”Supruniuk says the European defence sector should rethink the balance between quality and quantity. “In many cases, a good-enough system available in thousands of units creates greater operational value than a perfect system available in dozens.”“For decades, everyone has been looking for that all-powerful, exquisite system that can protect everyone,” adds Tymchenko. “We see in crime that only scalable, dedicated, affordable solutions can actually defend democracies.”
Tuesday Scaling the Rockies: Hartford Yard Goats begin playoffs tonight
Colorado Rockies news and links for Tuesday, September 15, 2026
Monday Scaling the Rocks: Country Roads, Take Me Home
Colorado Rockies news and links for Monday, September 14th, 2026
Unhinged Business Co. Receives Best Scaling Start Ups Advisory in Australia of 2026 Recognition for Commercial Strategy and Founder Focused Growth
Unhinged Business Co. receives 2026 recognition for its commercially focused approach to helping founders navigate business growth and scale.Mel...
Core Growth Group Releases Plumbing Business Growth Strategies Guide for 2026
Core Growth Group is addressing 2026 growth pressures in the plumbing sector with new guidance on scaling beyond revenue gains, as workforce dem...
Sony is reportedly preparing a big update for the PS5 and PS5 Pro with improvements to PSSR 2.0 upscaling
Sony is reportedly getting ready to release a major new system update for the PS5, with firmware version 14.0 currently is beta.
DLSS 5 was launched in Medal of Honor European Assault using the PCSX2 emulator on an RTX 4070 Super.
On Reddit, enthusiasts demonstrated the DLSS 5 upscaling technology in the classic shooter Medal of Honor: European Assault,...
Big Ideas Lab podcast takes on the science of scale-up
Rechargeable lithium batteries trace back to 1972, but they wouldn't reach consumer products until 1991. Solar photovoltaic cells existed in the 1950s, long before solar panels appeared on rooftops. And the MRI was demonstrated in the 1970s, years before the machines became hospital staples. In each case, the core scientific principles had been proven. The difficult part was scaling them into technologies people could reliably use. The average disruptive technology takes roughly 35 years to cross that divide. Lawrence Livermore National Laboratory (LLNL) scientists are trying to compress
Colorado Sues EarnIn For “Immoral, Unethical, Oppressive and Unscrupulous” Business Practices
Hey all, Jason here. When this hits your inbox, I should be en route to the airport — not to go anywhere myself, just to pick up my better half. I must admit, it’s been strange to be home alone (well, with two dogs) for two weeks! Really lets me “optimize” to spending more time on reading/writing/working, whether that’s healthy or not! Looking forward to getting back into a (somewhat) more balance routine. Subscribe or Support by Upgrading Compare Notes With the People Who Built Modern Fintech Partner content: Remitly CEO Sebastian Gunningham. Increase founder Darragh Buckley. Linda Du, Co-founder & President at Valon. They’ll unpack the decisions behind building and scaling major financial products: what they got right, what broke, what they would do differently, and how AI and new infrastructure are changing the operator playbook. Most events tell you where fintech is going. NerdCon puts the people who built the present in the room and asks them what comes next. Expect an operator-led conversation grounded in real products, real trade-offs and lessons earned the hard way. Fintech Business Weekly readers save 20% on Regular tickets with code FBW20 before standard pricing ends September 11. Meet the Builders Things To Know & Other Good Reads Advent and Stripe Abandon $50 Billion Pursuit of PayPal (Bloomberg) Trump Family’s New Crypto Bank Is Backed by Abu Dhabi Sheikh (Wall Street Journal) The multiplying risks of financing data centres (FT) Let the Bond Market Speak (Wall Street Journal) Did we waste a crisis? Modest proposals to reform deposit insurance. (Fintech Takes Banking) Untangling Guggenheim: How Private Credit Built Its Own Universe (Net Interest) The Hater’s Guide To Circular Financing — Part One (Ed Zitron) Fiat Ventures Rebrands to FGV Capital, Announces $35M Oversubscribed Fund II (BusinessWire) Socure Announces Strategic Growth Investment at $5.2B Valuation and Acquires Agentic Operations Platform Fravity (Socure) Listen: When Consumer Protection Disclosures Work Too Well (Consumer Finance Monitor Podcast) Colorado Sues EarnIn For “Immoral, Unethical, Oppressive and Unscrupulous” Business Practices When Andreessen Horowitz led EarnIn’s $39 million Series B in 2017, Alex Rampell, a partner at the storied venture capital firm, described EarnIn’s small-dollar advance product as “free,” relying on “entirely on voluntary contributions instead of fees.” Rampell favorably compared EarnIn’s approach to typical payday loans, writing, “Payday loans historically have had the potential to be a slippery slope for consumers into financial distress: opaque systems and steep fees that are hard to repay and set individuals back further than where they started. For many Americans living paycheck to paycheck, that kind of slide into debt can be extremely difficult to recover from.” But EarnIn’s product — which, the company says, isn’t a “loan,” legally speaking — ends up being anything but free for most users. While it is technically possible to take a no-fee advance from the company, most users incur “Lightning Speed” (instant funding) fees and/or ostensibly optional tips that combined can equate to annualized percent rates that can reach over 1,000%. With an average term of 9.77 days, even fees and tips that are small on an absolute dollar basis are equivalent to high APRs on an annualized basis. A lawsuit filed last week by Colorado Attorney General Phil Weiser against EarnIn describes the company and its product and business practices quite differently than Andreessen’s Rampell; “EarnIn’s business practices are unfair because they are immoral, unethical, oppressive and unscrupulous,” the complaint, filed in District Court in Denver, Colorado, says. EarnIn’s marquee product, Cash Out, isn’t a loan, the company says, as Cash Out is not “[f]orward [l]ooking” because “[t]he money is transferred based on earnings to date,” does not carry a mandatory fee to access funds, and is “non-recourse,” meaning EarnIn cannot pursue recovery from users who don’t repay advanced funds. EarnIn is sometimes referred to as an “Earned Wage Advance” service. Though, unlike some others in the broader category, EarnIn does not integrate with employers’ payroll or time and attendance systems. Rather, EarnIn attempts to verify a user’s income and employment, by examining direct deposit transaction data in their linked external bank account, by using GPS location and Bluetooth data to estimate a user’s time spent at work, if they work in a fixed location, and/or by having a user provided a work email address. While repayment is theoretically optional, as EarnIn would have no recourse, such as reporting non-payment to the credit bureaus or a collections agency, this is anything but clear to users of the EarnIn app, Colorado’s complaint argues. During the process of taking a Cash Out, users must agree to a preauthorized debit of the amount advanced, plus any expedited funding fee and/or tips. Though users technically can revoke this ACH authorization, the fact that they have this right isn’t presented to users during the Cash Out process, but rather is included in linked terms and conditions few users are likely to actually read. And users who do wish to revoke their ACH authorization must do so three or more days before the scheduled transaction date by emailing EarnIn’s customer support. Similarly, EarnIn’s marketing claims that it carries “no interest” and “no mandatory fees” is technically true, but is not the reality for many of the app’s users, the Colorado complaint argues. Further, EarnIn has extensively marketed to users the ability to “access your pay today,” to get “instant” funds, and to get funds “the same day you work” — when, in reality, this speed of access was only possible by paying the additional, undisclosed Lightning Speed fee. The Colorado complaint highlights this contradiction specifically, noting: EarnIn tells consumers that they can ‘Make any day payday with EarnIn’ and that there no ‘hidden’ or ‘mandatory fees.’ For a consumer to access their pay on the same day requested, they would have to pay the Lightning Speed fee. In the fine print of the advertisement EarnIn discloses that ‘[f]ees apply to use Lightning Speed,’ a direct contradiction to the larger text of the advertisement, and a fact that would not be immediately understood by a consumer as ‘Lightning Speed’ is not defined nor explained. Like other small-dollar lending services, rather than explicitly charge interest, EarnIn collects a markup on the optional Lighting Speed fee, which, given the nature of the transactions, many users opt for, as well as offering users the option to leave “tips.” EarnIn leverages either Real-Time Payments (RTP) rails or push-to-card via debit rails if a user opts for the optional Lightning Speed. But while these capabilities cost EarnIn, on average, $0.075 or $0.20 per transaction, respectively, EarnIn charges users significantly more. Initially, EarnIn charged as little as $1.99 for Cash Outs up to $24. But EarnIn has hiked these fees multiple times, and now charges $4.99 for Cash Outs up to $75 and $6.99 for those over $75. According to the Colorado complaint, EarnIn leverages “dark patterns” to manipulate users and to make it more difficult to avoid tipping. Examples of these “road blocks” highlighted in the complaint include: As of 2023, in the EarnIn app, a consumer attempting to get a loan under the default settings had to make at least eighteen separate taps to complete the transaction and reduce the tip to $0. There is an alternative method to leave a $0 tip, via the “Custom tip” feature, but EarnIn made this option difficult to access by visually deemphasizing it and positioning it near devices’ “home” button. Using language and imagery such as “Tip to pay it forward,” “Your generosity supports the service + helps,” “Your tips make a difference,” and “[t]he APR for this cash out is 0%. Your tips help support us.” While Lightning Speed and tipping are, in theory, both optional, in practice, users nearly always ended up paying something. According to the Colorado suit, users were charged either a tip or an expedite fee for more than 92% of transactions, with an average APR of nearly 388%. And, although the advances are “non-recourse,” EarnIn successfully collects on nearly all transactions — 99.18%, according to the complaint, which, Colorado argues, “demonstrat[es] that EarnIn’s advances function as high cost loans, not voluntary payments.” The share of EarnIn users incurring fees — despite the product being marketed as 0% APR — is hard to reconcile with Andreessen investor Rampell’s favorable description of EarnIn vs. “opaque systems and steep fees” associated with payday loans, which, Rampell says, can be “extremely difficult” for borrowers to extricate themselves from. According to Colorado’s suit, EarnIn’s Cash Out product “trapped many consumers in extreme cycles of high-cost reborrowing.” The suit gives an example of one consumer in the state, who took out a whopping 1,151 loans, paying a total of $4,038.50 in Lightning Speed expedited funding fees on loans that averaged the equivalent of 1,421% APR. Another Colorado user, the suit says, took out 1,033 loans, paying $8,561.22 in tips and fees at an average APR of 1,539% . These types of usage patterns demonstrate “the severe and repeated financial harm caused by EarnIn’s illegal lending model,” the Colorado complaint argues. In aggregate, from January 2023 through July 2025, EarnIn extended 3.1 million loans to 56,778 users in Colorado — an average of nearly 55 transactions per person during the time period. EarnIn lent more than $300 million and collected more than $16 million from Colorado users in Lightning Speed fees and in tips. Ultimately, the Colorado complaint argues that “EarnIn’s Cash Outs are loans under Colorado law, and EarnIn’s contention that its product is not a loan based on its disclaimer of any legal repayment obligation lacks any real-world significance given how the Cash Out transactions actually operate in practice.” Colorado further argues that “EarnIn was not working with companies here in providing consumers with funds but acted as a third-party lender and charged illegally high rates, used deceptive design strategies to extract some charges, and trapped consumers in repeat borrowing. Colorado will continue to stand up for consumers and hold companies accountable when they violate our credit laws or attempt to evade them through misleading practices.” Colorado’s suit alleges EarnIn assesses finance charge in excess of that permitted by Colorado law, that EarnIn failed to make legally required disclosures to Colorado borrowers, that EarnIn made supervised loans to consumers without obtaining the requisite license, that EarnIn violated the state’s Deferred Deposit Loan Act, that EarnIn engage in unfair and deceptive trade practices, and that EarnIn made false or misleading statements concerning price. The suit seeks a court order enjoining EarnIn from violating relevant Colorado laws, for EarnIn to refund amounts charged in excess of state law, fees, costs, and penalties, and other relief the court deems to be just. Fintech Business Weekly is made possible by the generous support of paying subscribers — bringing you independent analysis of banking, fintech, and crypto without fear or favor. You can support my work by becoming a paying subscriber if you aren’t already. Paying subscribers enjoy: access to the entire archive of nearly six years of newsletters extended versions of the weekly newsletter, with additional content and analysis and (for founding member tier) quarterly personal 1:1 fintech Q&A / strategy calls with me, tapping into my unparalleled knowledge of the intersection of banking and fintech and experience working in the sector, including helping to launch Goldman Sachs’ retail bank Marcus Support Fintech Business Weekly You can also support Fintech Business Weekly by sponsoring a newsletter or podcast, putting you in front of 93,000+ of the most influential decision makers in banking, fintech, and crypto. Learn more about sponsor opportunities or request a media kit by dropping me an email. Q2 Quarterly Banking Profile: Five Quick Takeaways The FDIC released its quarterly banking profile, which is always worth a read, for the second quarter. The profile provides an aggregate snapshot about the overall health of the 4,238 institutions whose deposits are insured by the FDIC, who, collectively, hold more than $26 trillion in assets and saw a total net income of more than $90 billion in the second quarter. Net income was up 12% vs. Q1 2026, with lower provisions and realized gains on securities helping to drive the improvement, somewhat offset by increases in non-interest expense and applicable income taxes. Overall net interest margin was basically flat quarter over quarter (up 1 basis point vs. Q1 2026), though community banks saw NIM grow by 10 basis points, to 3.81%. Unsurprisingly, given that interest rates haven’t changed, unrealized losses on securities changed little, edging up slightly to $326.7 billion vs. $325.1 billion in Q1 2026. Aggregate credit quality has remained benign, with asset quality actually improving slightly in Q2. Overall, institutions reported a 9 basis point decline in their past-due and non-accrual rate, to 1.44%. Net charge off rate also declined by 2 basis points, to 0.57%. And finally, the FDIC’s deposit insurance fund stands at $161.1 billion, making the DIF reserve ratio 1.48% — above the statutory minimum of 1.35%, but still below the designated target reserve ratio of 2%. [Paid Subscriber Exclusive] Nissan Withdraws ILC App, State Associations Plan BankChain Alliance, FDIC Defines “Unsafe or Unsound Practice” For First Time Automaker Nissan withdrew its application to charter a Utah industrial loan company and its corresponding application to the FDIC for deposit insurance last week, on August 19th. Nissan’s application had been outstanding for more than a year, as it originally applied in June 2025. Recent applications to the OCC to form national banks have been decisioned, one way or the other, substantially more quickly than has historically been the case.
NASA’s Swift Restarts Science Observations
NASA’s Neil Gehrels Swift Observatory has resumed operating two of its three science instruments, following the scaling back of a commercial mission that
Gladys Knight sparks concerns, cuts back performances
Gladys Knight is scaling back her performance schedule after recent concerts prompted concern about the 82-year-old singer's health.
UFC fighters fined for climbing octagon fence after victories
UFC fighters Max Holloway, Paddy Pimblett and Gable Steveson were fined $2,500 each for scaling the octagon fence at UFC 329 on July 11 in Las Vegas.
Want to Scale Your Business? Stop Scaling Payroll First
Small businesses should build better systems to increase capacity and grow revenue faster than payroll without sacrificing employees.
Disney Tightens Spousal Health Benefits as Industry Costs Surge
The entertainment giant joins a growing list of major employers scaling back coverage amid a projected 11 percent rise in 2027 healthcare costs.
Child Sexual Abuse Is Preventable | Johns Hopkins Bloomberg School of Public Health
We could end CSA by focusing on solutions that work—and scaling them at a national level.
The Blood of Dawnwalker final PC specs are much lower, but upscaling is now recommended
The Blood of Dawnwalker now lists an RTX 4070 Ti for 4K 60 FPS Ultra with upscaling. Its earlier native 4K requirements called for an RTX 5090.
Enterprise AI Is Scaling Fastest Where Businesses Can Measure the Results
For all the attention paid to enterprise artificial intelligence strategy, adoption is not spreading evenly across the organization. According to findings
14 major artists who canceled or postponed 2026 tour dates
The Beach Boys, Rod Stewart, and Brandi Carlile are among the many acts scaling back their 2026 tours. Here are the biggest cancellations and postponements so far.
One-of-a-kind business in Champaign celebrates first anniversary
CHAMPAIGN, Ill. (WCIA) — A unique business in Champaign reached new heights with its first anniversary on Sunday. Horizon Climbing was packed with people celebrating, scaling rock-walls and t…
Growing Into the Middle Market: How Small Businesses Size Up
National Business Capital's Joe Camberato is back on the podcast to discuss scaling small businesses with a calibrated capital stack
The 5 AI Scaling Mistakes That Could Derail Your Business
Here are five costly mistakes businesses make when scaling AI, from runaway costs and weak governance to poor accountability plus how leaders can avoid them.
Oblivion Remastered on Switch 2 Outshines Xbox Series S, as NVIDIA DLSS Becomes The System’s Secret Weapon Again
The Elder Scrolls IV: Oblivion Remastered on Switch 2 tops Xbox Series S image quality via NVIDIA DLSS upscaling
CME venture with FanDuel falters as prediction startups race ahead
CME Group Inc. and FanDuel are scaling back a joint effort to take on prediction market startups after failing to gain traction.
MrBeast Sales Slow, Angel Stock Falls, MUBI and Neon Look for Hits, and Will an NBA Franchise Set a Sales Record?
(Welcome to the Entertainment Strategy Guy, a newsletter on the entertainment industry and business strategy. I write a weekly Streaming Ratings Report and a bi-weekly strategy column, along with occasional deep dives into other topics, like today’s article. Please subscribe.) You probably remember egg prices spiking a few years ago. This led to a lot of debate between the pro-market/pro-competition/antitrust/Neo-Brandesian crowd and the center/center-right/center-left/abundance/libertarian crowd over the cause, whether it was the avian flu outbreak or whether industry consolidation led to price fixing, using avian flu as a convenient cover. Basel Musharbash wrote a whole investigation at BIG looking at it. Well, two years later, we have our answer: it was price fixing. The Justice Department found proof, via emails, of price fixing. Indeed, just the threat of an investigation brought prices down two years ago: I bring this up because at least half a dozen of my favorite political writers and pundits, who were skeptical about price fixing at the time, haven’t written any mea culpas on this issue. It’s really dispiriting and makes it much, much harder to trust their work on other issues. If you take a hard stand on an issue, then you get it wrong, you need to explain how and why you got it wrong. Or update your priors. This is how epistemological bubbles are formed! All of which is to say, it’s time for another installment of “What I Got Right, What I Got Wrong”. Last issue, I mostly focused on the “wrong” side of the equation, so today’s is a bit more balanced, but with one big mea culpa (though I still stand by my original concerns). Let’s dive right in! Subscribe CORRECTION: UFC Didn’t Release Subscriber Data (But Seems to Have Bragged About It Internally) In my analysis of the viewership of UFC Freedom 250, I wrote... “Paramount has also released datecdotes bragging about their subscriber numbers. Earlier this year, they claimed that the first UFC fight on Paramount+ brought in 1 million subscribers.” Well, that datecdote did not come from a PR rep, but from great reporting from James Faris at Business Insider. The actual quote was this: “Paramount’s flagship streaming service generated about a million new subscribers on the day of its first-ever UFC event, Paramount product chief Dane Glasgow told employees in a town hall on Tuesday morning, three staffers who attended the meeting told Business Insider.“ This wasn’t the biggest goof in the world; even Faris emphasized to me that this is more a “clarification” than a correction. Still, it sounds like, internally, the folks at Paramount were very happy about this result—and I’m guessing they were happy with this news getting out—even though the gains didn’t last through the quarter, dropping to 700K adds. RIGHT: MrBeast’s Struggles Continue... I’m very skeptical about MrBeast’s burgeoning media “empire” and, in particular, its $5 billion valuation, as I wrote earlier this year and last year. Sure enough, according to Business Insider, regarding its “booming” consumer products division_:_ “US sales volume grew 13% year-over-year in 2025, following a 33% leap the previous year, according to a presentation deck dated May 2026....A February 2025 investor deck viewed by Business Insider showed Feastables’ net revenue more than doubled in 2024 from 2023, to $215 million, and was forecast to grow 74% in 2025, to $375 million.” [Emphasis mine.] If you told investors that you projected 74% growth but delivered 13%,1 that’s not ideal. But it’s not just MrBeast. Logan Paul and KSI’s Prime had sales drop in half for their energy drinks. It’s one thing to build a successful entertainment company and an entirely different thing to succeed in the heavily-consolidated-but-also-very-competitive consumer packaged goods space! CPGs are dominated by a few firms with very wide moats compared to YouTube, which has few barriers to entry. These businesses require different skill sets and have completely different business environments. WRONG: Does the NBA Have an $8 Billion Bidder? Almost as soon as I published my article about the NBA’s finances, casting skepticism on NBA expansion, I saw this Front Office Sports headline: One group “claims to have already raised $8 billion” for the Vegas expansion team. I mean, wow. That’s just $2 billion less than what the Los Angeles Lakers were valued at when they were sold just last year. And they’re arguably the most popular team in the NBA. (Side note: the new owner is now under an investigation for fraud in how he went about financing the Lakers and the Dodgers purchases.) So much for the “dried up” language I used in that article. (This is why I usually stick to my “Entertainment Nuance Guy” tone in articles...) I went off old information—from April, when I read/heard about the NBA downplaying expansion efforts—instead of doing a quick double-check for new news. That said...if you read the article, the market for Seattle is “slow”, with just one reported bidder. So I wasn’t entirely off. This goof has annoyed me since I first published it, mainly because it undercut that article’s real point, which I stand by: NBA franchises are really overvalued right now. Now, I get why NBA teams are overvalued. When it comes to owning sports teams, it’s basically never wrong to assume that some rich person somewhere will care more about owning a team (or just being part of a group since they’re so expensive) than the actual financials. There are only so many teams to own, and opportunities to buy in preferred markets don’t come along very often. This explains why the teams sell for such high multiples (usually above a 10 to 12.5 revenue multiple, which is very high). You only need one person to overpay! Beyond the ultra-wealthy luxury good explanation, I worry NBA teams are becoming speculative assets. More and more private equity investors are buying into NBA teams. But the value, in owning a team, comes from selling it later, not any meaningful relationship to revenue or profit growth. In other words, the newer NBA owners are increasingly expecting someone later to buy their inflated asset. That’s not sound economics; that’s the “greater fool” theory at work.2 At some point, actually making money has to matter. Especially since revenue growth is now below the growth of the stock market. (And this matters because the NBA doesn’t have easy levers to pull to drastically lower their costs—half of their revenue has to go to players and then you actually have to run a team, including coaching, training, travel, media expenses, and more—or raise revenues aside from one-off expansions in America and overseas.) I just don’t see how investors make their money back buying NBA teams, except for selling at a higher price later on down the line. But if revenue/profit growth continues to slow, at some point, someone will be left holding the bag. RIGHT: Baseball and Basketball are about equally popular. Related to the above, I wanted to cite two quick pieces of evidence for one of my other theories: that the NBA and MLB are about as popular as each other, but the NBA gets much, much better media coverage. One YouGov survey has that fans prefer basketball to baseball by 52% to 48%, which I’m guessing is within the margin of error for that poll. That’s virtually even, meaning they’re about the same popularity. Over on NBC, a recent Red Sox-Yankees game averaged 4 million viewers, higher than Sunday Night Basketball’s average of 3.4 million on the same network. A Yankees-Dodgers game averaged 3.9 million two weeks ago. Not every game averages this much, but it shows you that baseball isn’t far off from basketball, depending on the matchup. Altogether, the NBA is just slightly more popular than baseball, but NBC is paying MLB a fraction of what they pay the NBA, a point I’ve made before. RIGHT: Angel, MUBI and Neon Updates... I’ve long been skeptical about many of Hollywood’s buzziest startups/smaller indie studios. It’s not that they don’t make great movies—films like Anora and Everything Everywhere All at Once are amazing—or that they’re not (rightfully) pushing the majors to make better films, but...they get a ton of hype, and that hype often doesn’t live up to the reality or difficulty in scaling up. Starting with MUBI, well, The Wall Street Journal wrote a feature on MUBI’s struggles last April, best summarized by the subheader “After Sequoia valued Mubi at $1 billion, a left-wing revolt sent the indie film company into a tailspin”. MUBI lost $7.3 million on $200 million in revenue in Q4 2025, but reportedly hit a record 1.7 million subscribers at the end of Q1. As for Neon, Department M purchased a “significant stake” in the company earlier this year. How much is that “stake” worth? According to Deadline, “No financial figures for the Neon-Department M deal were disclosed.” So...that’s not great! If it was a big valuation, we’d hear about. Also, I found this curious: “Both Neon and Department M already have separate partnerships with Qatar. The Qatar Film Committee has a slate deal in place with the former, and a biopic production pact with the latter.” Color me skeptical about much Middle Eastern investment dollars. See: LIV Golf. Finally, as for Angel (formerly Angel Studios), their stock very, very slightly popped last month after Young Washington ($20 million budget) which made $45 million in the US, but their stock is still way down from its opening. Smaller Updates RIGHT: The WGA should focus on antitrust concerns! I’ve been on this for years. Post-WGA/AMTPT deal in 2023, I wrote an article for the Ankler calling on the WGA to focus and organize against industry consolidation. Sure enough, the WGA filed a lawsuit against the Paramount-WBD merger, and people are organizing around this issue now. I wish this energy had coalesced earlier (specifically to lobby Washington much earlier), but better late than never! WRONG: Actually, the chip shortage might be due to market manipulation. In my last “Long Reads for the Long Summer” article, I shared an article about how LLMs are causing memory chip prices to spike, which will make all electronics more expensive. Well, turns out, three RAM manufacturers control 90% of the market, and now a private lawsuit accuses them of price-fixing. So yeah. WRONG/UPDATE: The Odyssey will hold! Two weeks ago, I wrote, “As of this moment, we haven’t had a super-blockbuster, though The Odyssey, Spider-Man: Brand New Day and Avengers: Doomsday could all cross the $500 million mark. (At this point, The Odyssey would need a very strong hold to get there.)” Well, it’s safe to say it’s going to have that hold! It only dropped 30% in weekend two and IMAX screenings are sold out all month. Spider-Man: Brand New Day is also off to a great start. RIGHT: People are rushing to get shows out yearly. I’ve long been bullish on yearly, weekly-released shows, and Leslie Goldberg at The Ankler wrote about how many people in town are now listening. WRONG: Mark Carney reverses Canada’s stance on homegrown production. After writing about Canada installing domestic production quotas, Canada has backed off. WRONG: I don’t trust Spider-Man: Brand New Day’s online “views”. When I originally wrote about the latest Spider-Man film’s record-breaking trailer, I thought this was just YouTube views, but after watching this Corridor Crew video and doing more research, well, it’s all social views of the trailer globally. Really? That stat includes TikTok, Reels and YouTube Shorts views. Basically, I don’t trust it anymore. (This film will probably set record, but I trust the pre-sales data more.) RIGHT: Celebrity Production companies are still struggling. I’ve written about this for years, but the company that I didn’t track was Kevin Hart’s HartBeat, which was valued at $650 million but is struggling now and might just end soon, especially after Kevin Hart pulled his endorsement deals. And Bad Robot is moving to NYC. CORRECTION: SpongeBob is a “second run” title. In my 2025 recap article, I had _SpongeBob SquarePants_labelled as a “library” title, when it is still a “second run” show, with new episodes still coming out on Nickelodeon. CORRECTION: In my last “What I Got Right, What I Got Wrong” update, I offered a correction from Doug Creutz and misspelled TD Cowen. Thanks, Grammarly! 1 Though he could have increased prices to make up for the gap, he couldn’t have increased them that much. 2 There’s a theory, which I think that Daryl Morey created and Bill Simmons popularized, that, to be successful building your team, you only need a handful of bad GMs in the league to take advantage of. To be clear, Daryl Morey’s theory was also for signing bad contracts, that you could always get out of bad contracts. Basically, there’s always someone who will make a bad trade. It’s not the same exact thing, but similar enough to the Greater Fool Theory that I really like it.
Sony raises outlook after entertainment drives earnings jump
Sony has been shifting its resources toward accumulating entertainment assets spanning music, games and anime, while scaling back in low-margin areas in consumer electronics.
Eau Claire Outdoor retail closure highlights growing strain on downtown businesses
Another downtown Eau Claire business is scaling back and closing its storefront, marking what local business advocates say is a growing trend of economic hardship in the district.
New Podcast: Expanding Teen Mental Health First Aid in New York
New York is scaling Teen Mental Health First Aid statewide, empowering youth and communities with mental health literacy.
Your AI Strategy May Be Making Your Business Look Like Everyone Else's. Here's How to Ensure Customers Still Choose You.
As AI lowers the barriers to starting and scaling a business, creating a memorable brand and great customer experiences will become the ultimate competitive advantage.
AI-backed imaging workflow helps generalist radiologists perform like breast specialists
Scaling the solution across the U.S. has the potential to mitigate the current shortfall of breast radiologists, researchers contend.
MLOps transformation at BBVA: delivering business value faster
In a previous article, we explored how BBVA is scaling its Machine Learning capabilities with a new architecture based on AWS technologies.
A quarter-century of baseball, hoops — and scaling walls of ice? Denver Sports Commission celebrates 25 years
The group responsible for bringing sporting events to Colorado says there’s more fun on the horizon.
Bristol Myers Squibb Building Life Science Industry’s Most Advanced AI Factory on NVIDIA Vera Rubin
The pharmaceutical company has demonstrated how AI can transform how drugs discovery and is now scaling AI across R&D to accelerate the discovery of next-generation medicines.
Scaling the X-Labs Model is an Important Experiment at a Moment We Need Paths Forward
X-Labs seek to expand on what FROs have shown is possible: the generation of foundational infrastructure for entire new fields of research science.
What it Costs to Scale a Business
Most entrepreneurs think scaling is about accelerating growth, but the real challenge is managing complexity before weak systems, reactive decisions, and operational inefficiencies slow the business down.
Assassin's Creed Black Flag Resynced Performance Benchmark Review
Ubisoft's Assassin's Creed Black Flag Resynced breathes new life into the classic with modern graphics, gameplay improvements and full ray tracing support. In our performance review, we examine its graphical fidelity, VRAM usage, upscaling options, and how well it runs across a wide range of modern GPUs.
Another Detroit-area music venue struggles to keep its doors open
Three other Detroit venues closed at the end of June. This one says it's scaling back operations, starting in August.
Cognizant to scale to 5,000 Frontier Certified Engineers and 10,000 Frontier Business Operators
/PRNewswire/ -- Cognizant (Nasdaq: CTSH), a leading AI Builder and technology services provider, today announced it was committing to scaling its...
Shekel squeeze: How offshore talent is keeping Israel's startup nation scaling upwards
While reducing costs is one reason to hire offshore employees, ONE Offshore explained that most companies outsource to expand their teams.
The Next Chapter for Founder-Led Beauty Brands
At The Business of Beauty Global Forum 2026, Unilever Prestige CEO Mary Carmen Gasco-Buisson joined K18 founder and CEO Suveen Sahib and Tatcha founder Vicky Tsai to discuss scaling founder-led beauty brands.
Belkin launches new 14-in-1 Thunderbolt 5 dock with 140W charging and triple 4K display support
Belkin has launched a 14-in-1 Thunderbolt 5 Dock in China priced at 2,299 yuan (approximately $338). The Belkin dock ships with a 180W adapter and 1m cable and delivers 140W USB Power Delivery over a single Thunderbolt 5 connection, with bandwidth scaling from 80Gbps to 120Gbps for high-resolution display use. Belkin briefly listed the dock on its global website at $350 before removing the page, hinting at a global launch.
The 5-Question Org Chart Audit I Run Before Scaling AI Across a Business
AI success depends less on the technology and more on the organization behind it. These five questions help leaders identify the structural barriers that prevent AI from delivering real business impact.
Sephora CEO Guillaume Motte on Creating Demand and Scaling Desire
The beauty retailer’s chief executive spoke at day two of The Business of Beauty Global Forum about the strategies that have led to $5 billion in growth over the past three years.
America’s Friendship Recession Has Become a Startup Opportunity
A new “connection economy” is selling courses, clubs and confidence bootcamps to socially isolated adults, but scaling intimacy remains a challenge.
Why healthcare CIOs must fix their foundations before scaling AI
Healthcare leaders are focused on generative AI innovation, but the organizations most likely to succeed may be those investing first in architecture, governance and interoperability.
Pearl of the week: Scaling behavioral health integration
Learn how digital tools strengthen behavioral health integration.
AMD releases FSR 4.1 upscaling for older graphics cards
Now RX 7000-series graphics cards can use the upscaling tech.
Philips Foundation 2025 Annual Report: enabling access to healthcare for 69 million people by scaling what works
Philips Foundation’s 2025 Annual Report shows progress toward reaching 100 million people a year by 2030, with 69 million people gaining access to healthcare in 2025.
Is SMIC N+3’s Metal Pitch Smaller than Intel 18A’s?
Almost four years ago, we published that SMIC had started shipping 7 nm (N+1) chips. Now, SMIC is shipping its third-generation 7 nm (N+3) process in Huawei’s Kirin 9030, with a minimum metal pitch of 32.5 nm, about 10% tighter than the 36 nm minimum metal pitch shipping in Intel’s latest Panther Lake CPUs on 18A. The headline is true, but incomplete cherry picked metric. N+3 reaches the density of TSMC N6 through aggressive DUV multi-patterning and design-technology co-optimization (DTCO), but it pays for that in complexity, efficiency and process control. We found this and more in our reverse engineering and teardown where we cover SMIC’s N3 process technology, Huawei’s packaging, memory, architecture, and more. SemiAnalysis has been building a state-of-the-art teardown lab in Oregon capable of analyzing the world’s most advanced and important chips over the last year and half. We have already generated revenue on advanced datacenter chip teardowns including our recent reverse engineering of a major TSMC customer’s COUPE CPO optical engine + EIC 3D stack. This is the first public report from the SemiAnalysis Teardown Engineering & Evaluation Lab, or STEEL for short. The lab is aggressively scaling up and out and we’re excited to announce it publicly. This is a bit of inconvenient timing for TechInsights as they are private equity owned and currently being sold while having enjoyed virtually no credible competition for decades. This has led to TechInsights underinvesting in CAPEX. SemiAnalysis exceeds TechInsights in revenue despite no venture or private equity ownership and being founded only 6 years ago. Because we have no external investors and are founder led, we move faster, build faster, and we can release client chip teardowns for free regularly, while focusing on datacenter for our major clients. Here’s the first public image from our lab, the HiSilicon Kirin 9030 Pro SoC: This report will detail our teardown of the Kirin 9030 and our findings on SMIC’s N+3 process, the most advanced in China. For comparison, we’ll show our teardown of the MediaTek Helio G99, made on TSMC N6. Through this comparison, we can look at the effect of export controls – SMIC N+3 and TSMC N6 are comparable nodes, but one is heavily export-controlled, the other free to use the West’s most advanced equipment. Here we see both China’s progress and constraints. SMIC N+3 reaches TSMC N6-class logic density, but it requires far more aggressive DUV multi-patterning, so it does not match N6 on process maturity or cost. The Kirin 9030 Pro performs similarly to three-year-old Android flagships, and trails far behind the current flagship SoCs from Apple, Qualcomm, MediaTek, and Samsung. The efficiency gap is even wider. Export controls have not stopped Huawei and SMIC from shipping advanced silicon, but they have forced a different path. Without EUV, SMIC is leaning harder on DUV multi-patterning, DTCO, and increasingly complex integration. The roadmap continues forward through tighter design rules and backside power, but each step adds cost and process risk. Huawei’s τ scaling and LogicFolding show another path: stacking active logic and recovering density through advanced packaging and system-technology co-optimization (STCO). To understand the Kirin 9030, we must first understand Huawei’s SoC history. HiSilicon is Huawei’s chip design arm, responsible for the Kirin smartphone SoCs, Kunpeng server CPUs, Ascend AI accelerators, and switch/router networking silicon. Before export controls, Huawei was TSMC’s largest customer – the only customer on TSMC’s first EUV node, N7+, and among the first on N5, alongside Apple. That ended in late 2020. Huawei switched to Qualcomm SoCs in its flagship smartphones, though export controls limited them to 4G-only variants. In late 2023, Huawei returned to in-house silicon with the Kirin 9000s, a successor to the Kirin 9000, fabricated on SMIC N+2 instead of TSMC N5. In the following years, they released the Kirin 9010 and 9020 on the same N+2 process. These chips used Huawei’s in-house TaiShan CPU cores and Maleoon GPU. We have not torn down a Kirin 9020 ourselves, so the predecessor die shot is from Kurnal. The die shots show how Huawei spent its silicon budget: which functional blocks are where, and how their areas compare to the predecessor. First, a quick guide to the major blocks on the die. The total die area is nearly identical, but the 9030 uses that area more aggressively. A denser process lets Huawei fit an extra middle CPU core, more GPU and NPU cores, and larger caches into the same footprint. In contrast, the Helio G99 is a much smaller, low-cost SoC, built for budget smartphones rather than a flagship device. While the Kirin 9030 is ~140 mm², the G99 is only ~29 mm², roughly one-fifth the area. The underlying TSMC process technology, however, is directly comparable as a baseline for analyzing SMIC’s. The Kirin 9030 is an evolutionary refresh, not a clean-sheet design. Its CPU, GPU and NPU cores carry over the 9020’s families, and the gains come from three levers: the SMIC N+2-to-N+3 process step, DTCO and floorplan work, and incremental microarchitecture. Area is where the first two show up, and the 9030 scales well here. Performance and efficiency are the harder test. Huawei’s design holds up better than its node would suggest, but the chip still trails, both because N+3 sits behind the leading-edge nodes and because its cores, while competent, remain a few generations behind the newest designs. The new prime core is an incremental update. The main changes are a 10% frequency increase from 2.5 GHz to 2.75 GHz and a doubling in the L2 cache from 1 MiB to 2 MiB. Despite the increased cache, the core size decreased by 7.6%. Excluding the private L2 cache, the core size decreased by 21%. This is a large reduction for an incremental node. Compared with the TaiShan New V120 core in the Kirin 9020, the Kirin 9030’s middle core is almost unchanged architecturally, yet each core shrinks by ~22%. Most of that comes from the move from N+2 to N+3, with layout likely accounting for the rest. Visually, the most noticeable change is the increase from 3 middle cores to 4. There is also a 20% increase in the shared L3 cache of the big cluster. This helps to improve multi-core performance without sacrificing much in terms of area. Even with each core shrinking, the big CPU cluster’s total area is essentially unchanged. The per-core savings went back into an additional middle core and larger caches. The tiny cores shrank less than the prime core (excluding its L2 cache) and less than the middle cores. This is likely because fixed overhead is a larger share on a small core. We cannot resolve any architectural changes from the die shot alone, but the per-clock and efficiency gains shown below point to more than pure process and layout scaling. The area reduction was offset by a doubling of the shared L2 cache from 2 MiB to 4 MiB, leaving the total tiny CPU cluster area slightly larger. Area is the easiest improvement to see from a die shot, but it is only one part of PPA (power, performance, area). For modern logic, power and performance matter just as much, and often more. Since Dennard scaling broke down in the mid-2000s, voltage and frequency have not scaled in step with transistor dimensions, so each node has had to fight harder for gains in performance and efficiency. The starkest comparison is not Kirin 9020 versus Kirin 9030 Pro. Apple’s efficiency cores run circles around Huawei’s prime core. Apple’s low-power core delivers 20% higher integer performance while drawing only 1 W, compared with 4.5 W for Huawei’s prime core. N+3 matches TSMC N6, but N6 is several generations old. Apple and Qualcomm build on N4 and N3P, which are denser and sit on a better voltage-frequency curve, giving them a larger transistor budget and more performance per watt. The 9030’s own cores did improve. The middle and tiny cores gained 17% and 14% in per-clock integer performance over the 9020, with floating-point flat on the middle core and up 11% on the tiny. The tiny core improves cleanly, with performance rising while power falls and efficiency increases by 45% in integer and 24% in floating point. The middle core is mixed: integer performance rises but power rises faster, cutting integer efficiency by 7%, while lower power lifts floating-point efficiency 16%. Per-clock gains at the same or lower frequency are microarchitectural, so the cores are tuned, not just shrunk. Both also failed to hold their rated maximum frequencies, pointing to thermal, power, or stability limits. Per clock, the middle core sits around Arm Cortex-A720 and the tiny core near the Cortex-A520; absolute performance trails because Huawei clocks them much lower. The prime core is roughly Cortex-X2 class per clock, a 2021 design. Apple’s 2020 M1 Firestorm core is still 35% higher per clock and 57% faster in absolute integer performance at a similar 4.5 W. The current leading edge is further ahead again: the Apple M5 P-core is 60% higher per clock and 2.7× faster, the Arm C1 Ultra 45% higher and 2× faster. Matching older high-end cores per clock is a genuine design achievement. What Huawei cannot match is the voltage-frequency curve and transistor budget of leading-edge nodes, which let Apple, Qualcomm and others spend more transistors in the same area on wider cores, larger caches and deeper buffers while running at lower voltage. Huawei’s LogicFolding roadmap is one answer, stacking active logic to recover density and shorten signal paths. We return to it later. The GPU compute units (CUs) changed more visibly than the CPU cores, moving to a more rectangular layout for both the arithmetic logic unit (ALU) clusters and the CU overall. Even with ray-tracing support added, a CU shrank ~28%. However, that shrink is offset by the increase from 4 to 6 CUs and the area outside the CUs grew 33%. Overall, the GPU cluster is larger by ~10%. The GPU is where Huawei makes its biggest gains. The Maleoon 935 is not competitive with current flagships, but it is a large step up from the 920 and reaches older-flagship territory. In 3DMark it is 70% faster in Wild Life Extreme (WLE) and 79% faster in Steel Nomad Light (SNL) than the 920; with 11% higher clocks and 50% more CUs, the ~67% theoretical uplift roughly matches WLE and is beaten by SNL. It edges ahead of the Snapdragon 8+ Gen 1 in WLE and SNL, and the Dimensity 9200 and Apple A16 in WLE, but stays far behind newer parts: the Snapdragon 8 Elite Gen 5 and Dimensity 9500 are ~2.4–2.6× faster in WLE and ~3.2× faster in SNL. The Maleoon 935 is Huawei’s first GPU with hardware-accelerated ray tracing; there it lands slightly ahead of the Exynos 2200, and on par with the Apple A16, with current flagships up to 3.7× faster. The Neural Processing Unit (NPU) saw the largest structural changes of any block, moving from a Lite and a Tiny core in the Kirin 9020 to a Lite and two Tiny cores in the Kirin 9030. Both core types also show significant layout changes. This is a reversal in Huawei’s NPU design. The Kirin 9000 5G, its last flagship chip on TSMC N5, used two Lite and one Tiny core. The series of SoCs on SMIC N+2 moved to one Lite and one Tiny core, likely for area savings. With the Kirin 9030, Huawei has shifted back toward a larger multi-core NPU cluster, but with the additional area going to a Tiny core rather than a Lite core. We’re diving deep into the most advanced datacenter and AI hardware hitting the market. To learn more about what’s in the pipeline or to commission a custom teardown, contact sales@semianalysis.com. Interested in joining us on this ride and think you can be a difference maker? Check out our Careers page. Before diving into the process stack, the package and memory are worth separating from the SoC itself. The Pro variant of the Kirin 9030 carries 12 GB of Samsung DRAM, with two stacks of four dies each. The dies were identified as the K4L2E165YD, a 12 Gb LPDDR5X-9600 device fabricated on Samsung’s 1a node, the fourth generation of its 10 nm-class DRAM after 1x, 1y and 1z. 1a has shipped in volume since 2022, so this is current memory rather than older-node inventory. The 16 GB Pro Max variants we obtained were found with both CXMT and Samsung packages. The CXMT package is marked CXDD7JEDM, with two stacks of four dies, packaged in week 45 of 2025. The inferred die dimensions from X-ray computed tomography (CT) are consistent with a known density of ~0.3 Gib/mm² for the CXMT G4 process, roughly equivalent to other manufacturers’ 1z processes. The Kirin 9030 uses a typical integrated package-on-package (iPoP) stack: multiple DRAM dies in a memory package sit above an organic redistribution layer (RDL) interposer, which sits above the SoC and package substrate. The full package is then mounted to the printed circuit board (PCB) through ball-grid array (BGA) solder bumps. The memory package substrate is a thin bismaleimide-triazine (BT) laminate carrying the LPDDR5X stack. The organic RDL interposer over the SoC routes the PoP signals around the die and carries possible dummy thermal copper pillars. The package substrate, a thicker Ajinomoto Build-up Film (ABF) build-up over a BT core, fans the flip-chip bumps out to BGA pitch and embeds the power planes. The whole stack is organic. The only silicon is the SoC and the LPDDR5X dies; there is no silicon interposer. Keeping it all-organic brings the package’s coefficient of thermal expansion (CTE) close to the PCB’s, reducing board-level warpage, and avoids the cost of a silicon interposer the SoC’s bandwidth does not need. In an iPoP stack, the memory package connects to the organic RDL interposer through an array of solder bumps. Underfill fills the gap around those bumps, adding stiffness and protecting the joints from mechanical stress. The Pro and Pro Max variants differ here, which we cover behind the paywall. The die shot and architecture tell us how Huawei allocated its silicon budget. The process tells us what SMIC can manufacture. We use the Helio G99 as the process reference for TSMC N6. Both SMIC N+3 and TSMC N6 are evolutions of previous 7 nm-class nodes. We used targeted TEM cross-sections through logic and memory regions, imaged in both fin-cut and gate-cut directions. Each cross-section caption gives its horizontal field width (HFW), the real width of the imaged area. We start at the transistor fins, then move up through standard cells, local interconnect, and SRAM. SMIC has not overtaken Intel or TSMC. It uses aggressive DUV scaling and DTCO to reach N6-class density, but that density doesn’t translate into comparable performance and efficiency, for two reasons: the node gap to leading-edge nodes, and Huawei’s core designs. Fin Profile One of the most important knobs in a FinFET process is the fin profile: the shape of an individual fin and the channel where current passes from source to drain. The ideal fin is tall, narrow, and nearly vertical. A taller fin increases effective channel width, while a narrower fin improves electrostatic control by thinning the body the gate must control. Push either too far, and the process pays for it: weaker drive current, fragile fins, taper, footing and line-edge variation that hit yield and device variability. The Intel 22 nm, 14 nm, and 10 nm fin cross-sections show how FinFET nodes have improved over time. 22 nm fins were a first-generation structure, relatively short, wide and strongly tapered. The shape limits current density and reduces gate control uniformity across the height of the fin. At 14 nm and 10 nm, Intel pushed the fins taller and narrower while also making the sidewalls more vertical. Rather than shrinking the device, these changes increase the effective channel width per fin and improve electrostatic control. The trade-off is that taller fins at tighter pitches are much more difficult to manufacture. Now, let’s compare the Helio G99 on TSMC N6 with the Kirin 9030 on SMIC N+3. Both processes are in the same class, with fin pitch of 30-32 nm on N+3 and 34 nm in our N6 cross-section. The pitch for N6 is especially interesting as N7’s HD library is generally listed with a 33 nm fin pitch, and N6 did not shrink pitches directly. Its density gains came from DTCO instead of tighter pitches. The 34 nm pitch was stable across our sampled region and serves more as a comparison against the SMIC N+3 we have not investigated further. Pinning down N+3’s fin patterning scheme takes more than one core unit. The CPU cores show a dense ~32 nm pitch, with the pitch between N-P fin pairs alternating between 78 and 88 nm. Logic alone may be consistent with dual-pitch mandrels of 120 and 110 nm, but this is a complex and unusual approach. Combining the pitch from the the 8T SRAM, which has more complex repeat unit, with the CPU core sequence allows us to reverse engineer the patterning steps with more confidence. As both the logic and SRAM should share the same base grid, a single CD mandrel lithography pattern with 128 nm pitch undergoing SAQP produces a die-wide ~32 nm grid (128 nm/4), which supports the pitch sequencing seen in both logic and SRAM cells. In the sampled cross-sections, N+3 shows a taller, narrower, higher-aspect-ratio fin than N6. The measured fin aspect ratio is ~9.5:1 on N+3 versus 7.8:1 on N6. N+3 also shows less top rounding, with an estimated radius of ~2 nm, compared with 2.8 nm on N6. Even though the fin widths differ, the ratio of top rounding to fin width tells the same story, with N+3 at 0.37 and N6 at 0.44. In a geometric sense, lower is better; a perfectly rectangular fin would have no top-rounding penalty. These are single-digit-nanometer features measured from a handful of cuts, so treat the absolute numbers as approximate. The important result is the relative gap: N+3’s fins are consistently taller, narrower and less rounded than N6’s. We’re diving deep into the most advanced datacenter and AI hardware hitting the market. To learn more about what’s in the pipeline or to commission a custom teardown, contact sales@semianalysis.com. Interested in joining us on this ride and think you can be a difference maker? Check out our Careers page. Standard Cell A standard cell is the basic building block of chip layout: a fixed-height row pairing one NMOS and one PMOS transistor that share a gate, tiled in a grid to build logic blocks. The key dimensions are contacted gate pitch (CGP), cell height (CH), fin count, and the lower-metal routing grid. To measure density, we use the Bohr metric: a weighted average of NAND2 gate area (60%) and scan flip-flop area (40%). This represents a realistic mix of combinational and sequential logic. This metric has its limitations, especially for complex cell layouts like TSMC’s FinFLEX, which alternates cells with different fin counts. Even so, it is the best metric for a pure process-level comparison. Another important measurement is the fin pitch; it refers to the distance between two fins of the same transistor. In a FinFET process, multiple fins are used in each transistor to increase the drive current and thus performance. TSMC N6 ships both a high-density (HD) library with 2 PMOS and 2 NMOS fins per cell, and a high-performance (HP) library with 3 of each. More fins under the shared gate mean more effective channel width. HP cells switch harder at the cost of area. Designers mix the two on a die, primarily spending HP cells on timing-critical paths, and matching their PPA targets. In the Cortex-A55 core of the Helio G99, we found a cell height of 240 nm for the HD cell. MediaTek has used HD cells in the G99 to minimize die size and thus cost. As an SoC for budget smartphones at ~$100, this is essential. By contrast, we found only one library in the Kirin 9030, with 2 NMOS and 2 PMOS fins. This suggests a narrower library strategy than TSMC N6, where both HD and HP libraries are widely used. This likely reflects the smaller customer base and the more constrained domestic design and electronic design automation (EDA) ecosystem. In all three CPU cores of the Kirin 9030, we found cell heights of 228 nm, 5% smaller than on N6. This is also a reduction of 9.5% over SMIC N+2’s cell height of 252 nm. SMIC N+3 and TSMC N6’s HD library both feature a CGP of 57 nm. For SMIC, this is a 9.5% shrink over N+2. In the past, CGP and cell height alone may have been enough to compare transistor density. Now, however, we must consider scaling boosters and DTCO as well. SMIC’s density gain does not come from EUV. It comes from using every available DTCO booster aggressively. First is fin depopulation: reducing the number of NMOS and PMOS fins in each cell. The first FinFET nodes started with 3 or 4 fins for each transistor. SMIC N+3 and TSMC N6 HD both use only 2 fins per transistor, trading drive strength for density. Next is contact over active gate (COAG). By landing the gate contact directly over the active gate, instead of out over the isolation region, the cell height drops. N+3 integrates COAG while N6 does not. Our N+3 gate-cut cross-sections indicate COAG, with the gate contact sitting over the active region, while N6 shows an off-gate contact. Last is single diffusion break (SDB). Diffusion breaks are inserted between cells in the same row to provide electrical isolation, but they also introduce local layout effects (LLE), layout-dependent shifts in electrical characteristics. In the past, a double diffusion break was used, consuming the space of two CGPs. SMIC N+3 and TSMC N6 instead use SDB, saving area but increasing LLE sensitivity. This must be controlled at the process level and accurately modeled in the process design kit (PDK) so that EDA tools can account for it. Overall, SMIC N+3 has a transistor density of 113.4 MTr/mm², slightly above TSMC N6 at 107.7 MTr/mm². Even without EUV, SMIC has achieved density beyond TSMC’s mature N6 node which utilizes EUV. Metal Stack The smallest critical dimension in the teardown is M0; SMIC N+3 uses a 32.5 nm local metal pitch. That is smaller than the 36 nm M0 pitch on Intel 18A in Panther Lake. However, this does not mean that SMIC has a better process than Intel 18A or TSMC N3P. M0 is a local intra-cell routing layer. Its usefulness depends on the full interconnect stack: M1 and M2 pitch, track count, via and line resistance, design rules, mask count, overlay control, and routing flexibility. The 32.5 nm M0 is consistent with self-aligned quadruple patterning (SAQP), whose four-population line-width loading we read coarsely as alternating widths of 21.5 to 24 nm; M1 and M2, at 38 and 40 nm, are consistent with self-aligned double patterning (SADP), a single A/B split. On TSMC N6, M0, M2, and M3 sit at a relaxed ~40 nm and are consistent with SADP-class double patterning, with no need for quadruple patterning. That said, we measure M2 for example at ~43 nm, likely inflated by sparse routing. We do not assign any specific layer to EUV from our cross-sections; the distinction we can draw is double versus quadruple patterning, not lithography wavelength. Transistor-level density in the front-end-of-line (FEOL) sets an upper bound, but the design is ultimately limited by what the interconnect stack can route. The lowest metals are the most important for standard-cell density, but the semi-global and global layers determine how usable that density is at the block and chip level. Two axes are commonly used for chip cross-sections: the fin-cut and the gate-cut. The micrograph above is a fin-cut and shows metals 0 through 3. This axis lets us see the even-numbered metals, with M0 right above the fins. There are two kinds of M0 lines. The first are the power rails; these are wide wires for the VDD and VSS running horizontally at the top and bottom edges of each standard cell. The wide wires measure 55 nm across, more than double the other M0 lines. Their width minimizes resistance and reduces IR drop. The second kind are intra-cell wires, short segments within the cell that connect terminals to M1. These have alternating widths between 21.5 and 24 nm. The M0 pitch is 32.5 nm, a 19% reduction versus N+2 and N6. At this pitch, DUV patterning requires more aggressive multi-patterning, increasing mask count, overlay sensitivity, process complexity and cost. M0 is below what a single DUV-defined spacer (SADP) can resolve, so SMIC cascades a second spacer step (SAQP). The cross-section reflects the cost: the M0 trenches are visibly more re-entrant (narrower at the bottom than the top) than M1 or M2 on the same chip and carry a bright barrier-rich foot where the trench meets the etch-stop layer. That shape is partly the intended damascene profile, as a slightly narrow bottom helps void-free copper fill, but its magnitude at M0 is driven by the tight pitch and the higher trench aspect ratio. Intel 18A supports an M0 pitch of 32 nm, although Panther Lake has only shipped with a looser 36 nm pitch. This is due to Intel’s heavy usage of HP libraries. Among leading-edge nodes, 18A has the loosest M0 pitches due to PowerVia. With power routing moving to the backside, congestion is reduced, and the entire front-side metal stack can be used for signal routing. M2 is the first true inter-cell routing layer. It runs horizontally like M0 but spans across multiple cells to carry block-level signals. The M2 pitch sets the cell’s track height – the number of M2 tracks that fit between the VDD and VSS rails, defining what the library calls a 6-track or 7.5-track cell. This layer is the most important, limiting the routing of entire blocks. SMIC N+3 features a 5.7-track cell. The M2 pitch is 40 nm, a 5% decrease over N+2 and the same as N6. This shrink keeps the pitch at the edge of what is possible with double patterning. Future nodes will need to increase the number of masks for M2 as reducing the number of tracks is much harder due to the limitations in routing. The micrograph above is in the perpendicular direction, the gate-cut, and shows metals 0 through 4. This allows us to see and measure the odd-numbered vertical metal layers. The M1 pitch is 38 nm, 9.5% less than N+2 and 33% less than N6. The M1-to-gate ratio matters because it sets local routing flexibility. N+2 and N+3 use a 3:2 ratio, while N6 uses a 1:1 ratio, explaining the huge differences in M1 pitch. The more M1 lines there are compared with the gates, the more flexibility there is for power and signal crossing within the cell. Routing flexibility enables more complex and better cells. Clean fractional ratios are also preferred as a grid is periodic and improves layouts. The 3:2 ratio gives SMIC more local routing flexibility than a strict 1:1 grid, but it also complicates layout and patterning. This is a DTCO choice, with SMIC increasing process complexity to recover density and routability without EUV. This 3:2 ratio is not very popular in the leading-edge nodes. TSMC has only used it on N7+, the N5 family, and the short-lived N3(B). They have switched back to a 1:1 ratio for N3E. Intel has only used it on the 10 nm/Intel 7 family, with Intel 4, 3 and 18A all using a 1:1 ratio. Samsung is the only one still using a 3:2 ratio at the leading edge, using it in the SF4 and SF3 families. It remains to be seen if SMIC will remain at a 3:2 ratio or move to a 1:1 ratio with its future nodes. The industry is still actively exploring these local-routing ratios. At VLSI 2026, imec will be presenting work on even higher ratios, including a 2:1 scheme that can reduce area by up to 14%. We will be covering the conference in a future newsletter article. Subscribe We’re diving deep into the most advanced datacenter and AI hardware hitting the market. To learn more about what’s in the pipeline or to commission a custom teardown, contact sales@semianalysis.com. Interested in joining us on this ride and think you can be a difference maker? Check out our Careers page. The final local interconnect layer for N+3 is M3, with a pitch of 44 nm. The M3 pitch is the same as on N+2 and 10% larger than on N6. The semi-global layers carry the majority of block-level signal routing. They have a coarser pitch than the lower local layers. On leading-edge nodes, they are designed to sit at the limit of DUV single patterning. M4 through M11 pitches were found divided between 80–82 nm (M4–M6), 128 nm (M7–M10), and 148 nm (M11). Given limited sampling, it is possible these are divided further in dense routing areas. At the top are two giant metal layers, M12 and M13. These have kept the same pitches as N+2 at 1920 nm and 4600 nm respectively. While the lower layers’ pitches are generally fixed by the process and library, the upper layers vary much more in pitch and count, depending on the design. Even two smartphone SoCs on the same process can have wildly different metal stacks. The Helio G99 carries fewer routing layers, reaching coarse metal pitches of 850 nm by M9, while the larger and higher-performance Kirin 9030 keeps fine pitches until M11. SRAM At the leading edge, SRAM is much more difficult to scale than logic. TSMC’s latest nodes have seen little to no bitcell scaling, while logic still has more DTCO levers to pull. While looking for other logic libraries in the GPU compute units, we stumbled upon the SRAM. The most common type of SRAM has 6 transistors (6T), but this cell had 8 transistors (8T) instead. 8T SRAM adds two transistors to form a dedicated read port. Unlike a 6T cell, where reading disturbs the storage, the decoupled read port removes read-disturb, improving read stability and letting the cell be pushed harder for performance. At first glance, the cut looked like an unusual logic library, with each cell row having 3 fins of one polarity and 5 fins of another. The rows also alternated in orientation. Energy-dispersive X-ray spectroscopy (EDS) resolved our confusion. The cut had not landed on the GPU logic, but on the SRAM macro beside it. The unusual fin pattern was due to the SRAM library. We return to EDS in the process flow analysis behind the paywall. SRAM libraries are not like traditional logic libraries. Due to the unequal number of PMOS and NMOS transistors, they require specialized rules and layout libraries. They do not need the flexibility of logic libraries, so they are hyper-optimized for one purpose: dense, reliable memory. The SRAM cell we found is a 1:2:2-2:2 cell. This means there is 1 fin per pull-up (PU) PMOS transistor, and 2 fins per pull-down (PD) and pass-gate (PG) NMOS transistor. These 2 PU, 2 PD and 2 PG transistors would usually form a single 6T high-current cell (HCC). An 8T HCC adds a read-pull-down (RPD) and a read-pass-gate (RPG) NMOS transistor, each with two fins. We measured a cell height of 406 nm, which brings the bitcell size to 0.0463 µm². That is a theoretical peak density of 21.6 Mib/mm². We estimate that a 6T HCC would have a cell height of 292 nm and a size of 0.0337 µm². This is ~12% larger than a 6T HCC on Intel 3 and 4. We also estimate the 6T high-density cell (HDC) to have a cell height of 228 nm and a size of 0.0260 µm². This is coincidentally the same as the logic standard-cell height measured earlier. The estimate puts the cell near Samsung 7LPP/5LPP and slightly below TSMC N7/N6. That is a theoretical peak density of 38.5 Mib/mm². 6T HDC is arguably the most important cell as it is used for the largest caches in a chip, the L3 caches and system-level cache (SLC). Both the Kirin 9020 and 9030 have split the SLC into 4 banks to raise total SLC bandwidth. In the Kirin 9030, the SLC increased from 2 MiB to 3 MiB per bank. Correspondingly, the number of arrays within the bank also increased by 50%, from 16 to 24. Each array can store 128 KiB and forms an orderly pattern on the die shot. From the Kirin 9020 to the Kirin 9030, the area of a 128 KiB SLC array decreased from 0.0477 mm² to 0.0392 mm², an 18% shrink. The achieved density is 25.5 Mib/mm², 66% of the theoretical maximum. While the SLC was quite similar across both chips, the L3 has seen some major changes, particularly in terms of its layout. The total capacity also went up from 10 MiB to 12 MiB. Much like the SLC, the L3 is also split into 4 banks. In the Kirin 9020, an L3 bank consisted of 16× 128 KiB arrays and 16× 32 KiB arrays. However, an L3 bank in the Kirin 9030 instead consists of 48× 64 KiB arrays. In the Kirin 9020 L3, a 128 KiB array was 0.0513 mm² and a 32 KiB array was 0.0154 mm². The size of the 128 KiB array is different on the L3 and SLC as the assist circuitry for the two arrays differs depending on their purpose. In the Kirin 9030 L3, a 64 KiB array is 0.0210 mm². Although not a like-for-like comparison, normalized for capacity, it is 18% smaller than the 9020’s 128 KiB L3 array and 31% smaller than its 32 KiB L3 array. The achieved density is slightly lower than the SLC, at 23.8 Mib/mm², 62% of the theoretical maximum. Unlike the L3 and SLC, the prime cores’ private L2 cache uses a 2-bank design. As the prime cores’ L2 is latency-critical, it likely uses 6T HCC instead of 6T HDC. The 9020 has 16 arrays in each bank while the 9030 has 32. Each array has a capacity of 32 KiB. A 32 KiB array in the L2 shrank from 0.0171 mm² to 0.0142 mm², ~17% smaller. The density is 17.6 Mib/mm², ~59% of the theoretical maximum for 6T HCC. SRAM scaled well from N+2 to N+3, shrinking by ~19%, close to the theoretical logic shrink. The caveat is that N+2’s bitcells were unusually large, bigger than comparable 7 nm-class nodes, so part of the gain is catch-up rather than true scaling. With the insights from STEEL’s teardowns, we will be doing a deep dive into SRAM in a future newsletter article. Subscribe Everything above came out of a single STEEL teardown: die annotation, block-level area analysis and TEM cross-sections through logic and SRAM. We’re diving deep into the most advanced datacenter and AI hardware hitting the market. To learn more about what’s in the pipeline or to commission a custom teardown, contact sales@semianalysis.com. Future Roadmap The same cross-sections that pin down N+3 also show where SMIC can go next. Although N+3 is already close to the practical limits of DUV multi-patterning in several layers, SMIC still has a few scaling levers left. A theoretical N+4 would likely start with cell height. N+3 uses 5 M0 tracks between its power rails. Moving to 4 M0 tracks, as on SMIC N+2 and TSMC N6, could reduce cell height by roughly 15%. The routing grid is only one side of the shrink; the front end also must fit into the smaller cell. One possible FEOL lever is reducing the p-to-n isolation spacing from two diffusion grid units to one. Intel used this scaling booster on Intel 4, and TSMC did so on its N3 family. This path trades layout flexibility for density. Fewer M0 tracks reduce local routing resources, while tighter p-to-n spacing raises integration and design-rule difficulty. M2 is also constrained by the cell height shrink. For SMIC to maintain a ~5.7-track cell, M2 would need to move toward ~35 nm. That would move another layer into SAQP territory. SMIC could also reduce the CGP from 57 nm to 54 nm. Intel reached a similar CGP on Intel 10 nm/Intel 7 without EUV. The local interconnect is also tougher. If SMIC keeps the 3:2 M1-to-gate ratio, M1 would need to shrink to 36 nm and would likely require SAQP as well. If SMIC moves to a 1:1 ratio, M1 could relax to 54 nm, but it would give up routing flexibility. Under this theoretical path, we estimate that SMIC N+4 could reach a cell height of 198 nm and a CGP of 54 nm, implying a Bohr density of 137.8 MTr/mm², on par with TSMC N5 or Samsung SF4. However, the difficulty is cumulative. Each step is individually plausible, but together they make N+4 harder than the transition from N+2 to N+3. It will likely take longer, cost more, and carry less process margin. A theoretical N+5 would require a larger integration shift. One possible path is backside contacts (BSCon), moving power routing and source/drain contacts to the backside, which would reduce front-side routing pressure and enable another reduction in cell height. The front-side metal pitches could relax to reduce the process complexity. M0 would likely relax slightly to ~34 nm, and M2 and M4 pitches could relax further. CGP is unlikely to shrink much further. Even with EUV, 48 nm has been the practical limit for yield and process control. This approach would allow N+5’s cell height to fall to 170 nm and its CGP to 53 nm. This implies a Bohr density of 163.6 MTr/mm², on par with Intel 18A’s HP library. However, this would not make N+5 cost-competitive with the leading edge. It would reach a similar density through a much more expensive route. The integration difficulty rises sharply, with new process flows for backside alignment, wafer thinning, contact reveal, and backside metallization. Past this point, standard density and interconnect scaling become increasingly unattractive. That is where Huawei’s roadmap stops looking like a normal foundry roadmap and starts looking like a packaging roadmap. At ISCAS 2026, Huawei unveiled its tau (τ) scaling law, reframing process scaling in the time domain. τ is the time cost of data movement and processing: switching delays in transistors, RC signal propagation delays in circuits, compute, memory, and networking latency. Outside Huawei’s terminology, this is called system-technology co-optimization. This is Huawei’s answer to its lack of EUV lithography. Without EUV, planar density cannot keep pace with TSMC, Intel, or Samsung. If transistor density cannot shrink further, Huawei’s alternative is to shorten wires, reduce buffering, and stack logic vertically. “LogicFolding”, Huawei’s implementation of this new scaling idea, is, in practice, an aggressive 3D stacking approach. AMD V-Cache places SRAM above or below a CPU die. AMD’s MI350X places active interposer dies (AIDs) underneath accelerator and compute dies (XCDs), with the AIDs handling cache, IO interfaces, the network-on-chip (NoC) and embedded metal-insulator-metal (MIM) capacitors. With LogicFolding, parts of the same logic block are split across multiple active dies bonded face-to-face at ultra-fine pitch. This allows Huawei to shorten some critical paths and reduce buffer overhead, not merely add cache capacity or offload the IO and interconnect. Shortening wires is where the higher clocks come from. A large share of a modern core’s delay and energy budget goes into driving long interconnects and the repeater buffers along them. LogicFolding distributes a block’s critical-path gates across multiple stacked tiers bonded at a very fine pitch, so the bond interface behaves like an additional metal layer and the longest paths get shorter. That is how Huawei expects to recover frequency and efficiency it cannot get from the process alone. Huawei’s roadmap shows its intent. Prime core frequency is targeted to rise from 2.75 GHz in the Kirin 9030 to roughly 5 GHz by 2031, far beyond what planar scaling alone could deliver. Prime cores with 3.1 and 3.39 GHz clocks are being tested in its labs, although their power consumption is unknown. Beyond that, chips are in the design, simulation or pathfinding phase, meaning the frequencies are targets. However, the direction matters more: LogicFolding also helps with performance, not just density. The catch is that Huawei’s density claim is not directly comparable to foundry densities. A stacked design can report more transistors per package footprint by adding active layers, even if each patterned die remains well behind TSMC or Intel in front-end density. This is how Huawei can claim to reach foundry 14A-equivalent density by 2031. This is not a like-for-like foundry comparison, with Huawei using stacked logic and measuring density per package footprint. On a normalized Bohr-density basis, SMIC N+3 is ~114 MTr/mm², 38% less than Intel 18A’s HD library. Huawei’s 3D roadmap closes the gap by stacking active logic, reaching 215 MTr/mm² by 2030. In 2031, the roadmap density jumps to 295 MTr/mm², implying either a third active layer, partial EUV insertion or aggressive planar DUV scaling. Huawei’s methodology makes other foundries look much denser as well. Applying it to AMD’s MI450X with an N2 top die on an N3P base die yields a theoretical density of 460.2 MTr/mm² in 2026, compared with Huawei’s 295 MTr/mm² in 2031. This Kirin 9030 does not use LogicFolding, remaining in a conventional mobile SoC package. Instead, it forms the baseline for how far Huawei and SMIC can push planar scaling. Future teardowns of Kirin and Ascend chips will show both planar logic density and Huawei’s hybrid bonding solutions. Export controls changed China’s optimization problem rather than ending it. EUV restrictions raised the cost and complexity of leading-edge manufacturing without freezing it. SMIC reaches N6-class logic density through DUV immersion, SAQP and DTCO, while Huawei shifts more of the burden onto architecture, packaging and system-level integration. Future nodes will be tougher. N+3 still had room to tighten local metals and reduce cell height and CGP. Further scaling without EUV leaves fewer levers. More aggressive multi-patterning adds masks and overlay error. SMIC can keep pushing DUV, but each step will get more expensive and less forgiving. The design side is just as critical. Huawei had domestic EDA tools and flows before the Kirin 9030, with the Kirin 9000s, 9010 and 9020 making that clear. Huawei was able to ship multiple consumer SoCs on SMIC N+2 and N+3 while cut off from the Western EDA stack. US export controls restricted EDA tools for advanced chips in 2022 but did not target tools for more mature chips. In 2025, the US government briefly placed much broader restrictions on EDA software from Synopsys, Cadence, and others, before lifting them less than two months later as part of a trade deal tied to rare earths. Huawei has been unable to access those tools because it remains on a US trade blacklist. That forced Huawei, SMIC and Chinese academic institutions to build their own tools and flows. Researchers at Peking University recently announced a prototype EDA tool for Huawei’s LogicFolding architecture, which requires a new flow to handle the multilayer layout and floorplan. This is not the same as replacing the full Synopsys or Cadence stack, but it shows where domestic EDA is headed: toward tighter co-optimization between architecture, process and packaging. These advances are also diffusing into the Chinese ecosystem. SMIC is licensing its N+2 and N+3 processes to HLMC/Hua Hong at the government’s direction rather than by choice. If the same process learning feeds into Ascend accelerators for AI training and inference, the choke point shifts from one named fab to an ecosystem. Alibaba’s T-Head silicon arm and Cambricon, a Chinese AI chip designer that is expected to supply ByteDance, could also be major beneficiaries. Sanctions aimed at SMIC alone become less effective once the manufacturing knowledge has spread to other fabs and design houses. China is not closing the gap with Intel, Samsung and TSMC. The teardown shows the opposite in several places: no EUV, no backside power, higher process complexity, and visible trade-offs. But China is still advancing. If domestic chips become good enough for phones, inference, networking and security-sensitive workloads, they can matter strategically without matching TSMC at the leading edge. Behind the paywall, we show what else STEEL can do, with material and process flow analysis of SMIC N+3, and analysis of the Kirin 9030 package. We’re diving deep into the most advanced datacenter and AI hardware hitting the market. To learn more about what’s in the pipeline, access the full Kirin 9030 and SMIC N+3 analysis or to commission a custom teardown, contact sales@semianalysis.com.
My business is not my baby, and yours shouldn’t be either
This mindset is holding founders back from scaling.
WHO urges targeted action to strengthen emergency workforce capacities for shared health risks across the Asia-Pacific region
At a time when the world is more interconnected than ever, the World Health Organization (WHO) is disseminating a comprehensive regional analysis of shared health risks across Asia and the Pacific, urging Member States to invest in enhancing workforce capacities regionwide to mitigate these risks and ensure health systems are better prepared to anticipate, tackle and withstand a range of crises and shocks. Over 100 delegates from 49 countries and areas in the Asia-Pacific region reviewed results from the regional risk analysis during the WHO-led Asia Pacific Health Security Action Framework (APHSAF) Stakeholders Meeting in Malaysia. Synthesizing insights from 21 strategic risk assessments and more than 800 multisectoral experts, the regional analysis highlighted an urgent imperative: countries must proactively invest in emergency workforce preparedness to navigate a landscape where climate hazards, disease outbreaks and geophysical events increasingly overlap. These complex converging crises create devastating, compounding impacts on health systems and communities – reiterating the need for investing in a fit-for-purpose regional health workforce and multisectoral action.Key findings: an evolving risk landscape More than half of the participating countries ranked flooding, cyclones, dengue and pandemic-potential respiratory pathogens as being of high or very high risk, while antimicrobial resistance, landslides and gastroenteritis remain widespread concerns as well. “Given that 51% – over half – of these assessed hazards are deeply interconnected, isolated health interventions are no longer sufficient,” said Dr Saia Ma’u Piukala, WHO Regional Director for the Western Pacific. “Extreme weather events like cyclones and floods directly trigger subsequent disease outbreaks, while environmental degradation – driven by sea-level rise, coastal erosion and saltwater intrusion – fuels health risks across the Pacific. Mitigating these cascading risks requires unified, cross-sectoral action linking environmental, agricultural and public health policies to build true regional resilience.” Priority recommendations for action To address critical vulnerabilities, WHO calls on Member States in the Asia-Pacific region to urgently prioritize the following strategic actions: Bridge workforce capacity gaps. Embed and standardize emergency preparedness within academia to promote regional workforce interoperability. Foster multisectoral coordination. Address ad-hoc taskforce responses and fragmented surge support through formalized contingency plans and standard operating procedures that are frequently updated and tested. Secure sustainable financing for emergencies. Overcome systemic barriers to access dedicated funding for both immediate anticipatory actions and long-term emergency preparedness initiatives. Enable emergency-ready primary health care. Embed emergency management into public health workforce readiness planning. Ensure that emergency planning is inclusive. Establish clear pathways for inclusive emergency planning that actively involves communities most at risk and other vulnerable groups. Optimize communication. Update risk communication and community engagement strategies, aligning them with seasonal risk calendars. Elevating workforce readiness to meet regional health risks While the gaps identified are significant, the Asia-Pacific region possesses solid foundational pillars, including dedicated clinicians, laboratory staff, rapid response teams, emergency medical teams and extensive community networks. However, to adequately address the compounding nature of modern crises, the various components of the emergency workforce must collaborate cohesively. True resilience depends entirely on transforming this currently fragmented workforce into a highly trained, interconnected and rapidly deployable asset. To achieve this, WHO is actively advancing regional workforce capacities through the Global Health Emergency Corps initiative. This initiative – supported by the Gates Foundation and the Institute of Philanthropy – seeks to build a seamless network of emergency workforce across borders and sectors, ensuring that when a crisis strikes, a standardized, well-coordinated and trained workforce is immediately ready to respond. “Scaling up emergency workforce readiness is no longer optional,” said Dr Gina Samaan, Regional Emergency Director for WHO in the Western Pacific. “It is the definitive factor that will determine how effectively the Asia-Pacific region safeguards health and protects lives in an increasingly unpredictable future.” ____________________For more information, or to arrange media interviews, contact: Ara Johannes, WHO Western Pacific Health Emergencies Division: johannesa@who.int WHO Western Pacific Communications: wprocom@who.int
K-Pop Powerhouse SM Entertainment Elevates Global Hitmaking With Solid State Logic ORACLE
Seoul, South Korea, June 9, 2026 — SM Entertainment, the multinational record label, talent agency, production company and music publisher that pioneered the introduction of Korean music artists to | From shaping the sound of modern K-pop to scaling world-class production, SM adopts ORACLE to supercharge workflow, streamline sessions, and deliver signature SSL analogue excellence.
We can create the future of science right now
Paul Litvak is the founder and Executive Director of the Robyn Dawes Institute and a Visiting Scholar at UC Berkeley. He has a PhD in Behavioral Decision Research from Carnegie Mellon where he studied emotions and the sunk cost bias. Over 15 years in industry, he solved a wide range of challenging technical problems. At Meta he created models optimizing the ad review process. At Google he ran experiments to measure social influence. At Airbnb, he was a product manager leading machine learning teams optimizing search ranking and price suggestions. He also co-founded and led product at Rhythmic Health, a venture-backed biosensing startup, creating an accurate low cost system using a color changing strip and a smart phone to measure salivary lactate. The bottleneck At this point, it is uncontroversial to say that science needs to stop using the PDF article as the unit of knowledge and currency. The unbearable slowness of scientific publishing, the profit motive and margins: I’m not saying anything new. The PDF also sucks because it’s hard to extract structured information, which makes it hard to do evidence synthesis. As a result, we do much less evidence synthesis than is needed. And evidence synthesis ultimately undergirds most policy and medical decision-making. I can see second-by-second real-time odds for any sporting or newsworthy event, but a school board can’t see the best evidence on whether their 8th graders should be taught algebra. As a society, we don’t treat this as an important problem. Again, not controversial. Not only is the problem well understood, but the solution has already been laid out. What we need is AI-assisted living evidence synthesis - (1) an open knowledge graph of atomic claims, (2) claims linked to evidence, (3) assessment and synthesis of each piece of evidence, and (4) continuous updating with new data. What few realize (yet) is that the technical capacity to build this vision for a significant portion of science already exists. Not only that, scientists and startup teams are already building many of these components. I know this because I’ve been surveying the space and talking to many of the builders. There are some missing pieces: for example, evaluations of how well some of the components work. But at this point, most of what’s missing is a fully end-to-end working integration of all of these parts. In the rest of this essay, I’m going to lay out all the parts of a working living evidence layer for science and who is working on them, and propose concrete next steps for building this system. What’s now possible The diagram above outlines the components of a living evidence synthesis platform, including some of the teams working on each component1. Scientific PDFs are processed into claims with associated evidence. The evidence is subjected to a forensic audit, methodological evaluation and robustness and reproducibility checks. Finally it’s given a weight in a continuously updating synthesis. What follows is a description and status of each component and a few of the teams working on them. Document understanding The first thing you need to be able to do is turn an article into structured data. Mostly that means parsing PDFs. There are often multicolumn layouts that confuse non-specialized PDF to text processing libraries. For scientific papers, there is the added complexity of parsing formulas and tables and figures. This is a really hot area - there are startups offering APIs, and it seems like a new open source package gets posted to Github every few weeks. What follows isn’t exhaustive. A package called GROBID was the state of the art for a while, they didn’t update their package for nearly two years until very recently. In the meantime reducto.ai released an AI powered PDF extraction API, PaddleOCR became popular, IBM released a model called Docling, and both Mistral and Gemini created models and libraries. I also know of at least one other well-funded psychology research group working on a paper parser. By contrast, there are few open evals in this space, with no extensive evals for complex table comprehension in particular. Nonetheless, I’m confident this will be a solved problem soon, given the combination of LLM advances and developer interest. Hypothesis level extraction There has been increasing interest in comprehending the extracted text of papers and linking information to evidence for each hypothesis. A lot of work has already been done. Trialstreamer (Marshall et al. 2020) and RobotReviewer LIVE (Marshall et al. 2023) demonstrated automated extraction of trial population, intervention, and outcome at scale on clinical RCTs. PaperQA2 (Skarlinski et al. 2024) and Ai2 ScholarQA (2024) extended this to retrieval-augmented question answering with citation grounding. Elicit, Consensus, and SciSpace operationalized claim-level extraction for end users. OpenEval (Booeshaghi et al. 2026) is the most recent and most ambitious: 1.96 million atomic claims extracted from 16,087 eLife manuscripts using Claude Sonnet 4.5, grouped into ~299,000 results, with LLM evaluations showing 81% agreement with human peer review on a 2,487-paper subset. None of these solutions link claims to test-statistics, as you would need to evaluate randomized controlled trials. This is why I built the evidence.guide API - to extract hypotheses and associated test statistics from behavioral science papers. The best public eval of this kind of extraction I’m aware of comes from the recent SCORE project - they had humans code thousands of psychology papers to extract their claims by hand. It would be extremely helpful to the world if all scientific PDFs were available as structured open data. I’ve been working to make this happen, both directly at Berkeley and through coordination with large entities I can’t yet speak of; as hard as it is to do, I think it’s possible2. Forensic audit A lot of work has been done on forensic audit, but some gaps remain. Of course, for biology papers that rely on images for evidence, there are a variety of tools (notably Proofig and ImageTwin) to spot anomalies. These are still well short of what sleuths like Elizabeth Bik can do on her own, but these tools are constant companions among fraud analysts. There’s someone working on auditing Excel files for anomalies, and a number of teams are automating numerical checks like GRIM and SPRITE, including Scrutiny project, the INSPECT-SR team as well as statcheck. The regcheck team is building a way to use AI to compare preregistrations to analyses in papers, to ensure there aren’t significant deviations. Nonetheless, there are many other kinds of anomalies to screen for, both public and less publicly known. And there are no formal evals for anomaly detection that I’m aware of. Still, there’s a lot to draw from in this space and I’m pretty certain we will be able to scan papers for most kinds of obvious anomalies in the near future. Methodological review This area has been white hot, though I fear for many of the startups in this space, because this capability may become commoditized. There are at least six different AI peer review companies, including Refine.ink, Reviewer3, ReviewerZero.ai, Q.E.D. Science, Paper Wizard, and Isitcredible. Coarse (a pun on refine) was also recently created as an open source alternative. These systems provide qualitative feedback on the content of papers, spotting methodological weaknesses and mathematical errors. They seem to work pretty well, and many academics report bitterly that they exceed the average quality of typical peer reviewers. But there are few evals here either. What evals exist so far involve using LLM-as-judge (circularity problems abound) or comparing against human reviews of questionable quality. What you’d ideally want is an eval that measures capturing known errors in papers3. Reproducibility and robustness Another active area has been using AI agents to automate computational reproducibility4 and robustness5 checks in papers that report numerical results. For more recent papers where data and code are available, AI agents can see whether they can re-run the analyses and produce the numbers reported in the published paper. In addition to a handful of individual academics who have been experimenting using Claude Code for this, the Institute for Replication is a leading group working on building an end to end system. The evals related to this problem are the most mature, with CORE-Bench (Siegel, Kapoor, Narayanan 2024) and PaperBench (OpenAI 2025) available to benchmark agents on this task. There is also work on getting AI agents to test alternative ways of analyzing the data to ensure the results are robust to small analytic design choices. Synthesis This is the most underdeveloped area where significant investment is required. Although some automated evidence synthesis systems exist — for example, otto-sr is building an AI agent to write systematic reviews — none of these incorporate the full range of paper level signals to weight evidence appropriately. Nor is there anything like an eval or a gold standard for a good systematic review. Arguably Cochrane reviews are the closest we have to gold standard human systematic reviews, though I’ve heard academics in the know complain about their uneven quality. A key question for a synthesis platform is how to weight anomalies and methodological issues in assessing the quality of a piece of evidence. This is an unsolved problem and one I’m very keen to work on. Continuous updating There are many pieces of basic infrastructure available for monitoring for new research and initiating updates. OpenAlex is the current open citation graph. Retraction Watch integrated into Crossref in October 2023. Scite tracks how citations support, contrast, or mention prior claims. The Living Evidence Network demonstrated continuous-update workflows in clinical guidelines. Engineering this is a relatively straightforward task. When you look over this technical architecture and all the progress being made, it’s hard not to be optimistic that a living guide to scientific evidence will be built. The stakeholders are ready The social infrastructure for this is starting to coalesce — it’s not just a pie in the sky academic exercise to imagine this coming into existence. Institutions like the Center for Open Science, the Institute for Replication, the INSPECT-SR, the Living Evidence Network and more are all working on scaling work to improve research quality. Funders are also aligned. The Sloan Foundation has funded living evidence work through COS. Coefficient Giving supports the Institute for Replication and COS. The Astera Institute and the Institute for Progress have shown interest in this space. NIH has established an Office for Replication and Reproducibility. Although there are (very unfortunately) serious headwinds in science funding generally, there is an active group of funders interested in metascience. A brief word about what I’ve been doing at RDI. First, as a Visiting Scholar at Berkeley I’ve been actively figuring out how a non-profit and a public university can conduct and make public the results of large-scale academic article data mining. With some of the money I raised from donors, I commissioned a legal analysis of recent case law and publisher text data mining (TDM) agreements in order to understand whether a massive open data mining of academic articles is possible (with caveats, it is). I’ve also been working to bring together stakeholders in this space, and identify gaps. I’ve also been doing some software development in this space, with more to come. A pilot proposal The assumption undergirding all of this is that an AI, given all this information, would make the right judgment about a scientific claim with lots of conflicting evidence, weighing all the factors appropriately. That’s the hypothesis we need to test. Randomized control trial research is the best place to focus on first. RCTs are used to make many of the important decisions in society - from medical trials to public policy changes. And they use a relatively uniform set of inferential statistics with lots of known and available diagnostics. Behavioral science experiments, within the broader realm of RCTs, should be first used as a testbed whose results can be generalized. Because behavioral science is at the vanguard of open science practices, replications abound (there are thousands of them) to serve as ground truth training data. Key Hypothesis Therefore the pilot would test, in behavioral RCTs that have been replicated, whether the quality of evidence for a claim can be used to accurately predict whether that claim will replicate. Secondary Hypotheses Compared to claims that replicated, non-replicated claims demonstrate a greater share of forensic anomalies in their source literatures. Hypothesis level claim and statistic extraction is accurate enough to scale living evidence without onerous human review costs. Replication prediction is more accurate than prediction markets6 or journal prestige. If all the different quality signals we gather do accurately predict which studies will replicate, then we can use that model to score evidence to power the living evidence layer. Why this is informative regardless of outcome If the pilot succeeds, the architecture extends to medical RCTs (where Living Evidence already operates and integration is mostly about claim representation), then to slices of basic biology with stable replication structure. If it fails, the field learns which quality signals are load-bearing and which ones metascience has oversold. Either result is a contribution to knowing what the literature supports. Implications for funders Because this burgeoning ecosystem of builders already exists, a well-informed philanthropic or government funder could play a crucial catalyzing role in bringing this future about. They could play at least three roles: creating open structured datasets, publishing open benchmarks and incentivizing the solving of key technical challenges. First, open archives of papers that the government maintains, like PubMed, could be turned into structured data amenable to large scale metascience and claim aggregation. I know the US government already has an interest in doing this, though some key open questions remain unanswered. How do you determine the best models and systems for accurately extracting information from papers? How can you establish a robust way to allow researchers to flag errors and correct them? And finally, how do you create a legal regime in working with publishers to maximize the scope of available papers? For the latter, a university or a private philanthropy may be better positioned to make structured data publicly available under journal subscription terms or fair use. Philanthropists or government funders could also coordinate to create or commission benchmarks that evaluate whether important problems have been solved. For example, an open benchmark for claim extraction from a range of different scientific article types would be extremely helpful. Ensuring the underlying data are accurate is vital for creating these evaluations. I’ve discussed opportunistically using various human-created datasets for this purpose, but a consistent problem is that errors in human data make it difficult for them to serve as a gold standard. Finally, with structured data and benchmarks available, the government or private philanthropy could use them to incentivize groups to develop machine learning systems that meet these benchmarks. Prizes are one potentially valuable tool for this. For example, you could establish a prize for a system that accurately updates a living meta-analysis for a small set of claims. Prizes are particularly useful as signals of problem importance, and can help create vibrant ecosystems of public and private research—see, for example the role the government played in kickstarting the current work on nuclear fusion. Conclusion The drawbacks of the current scientific publishing system are known. Scientists agree, metascientists agree, philanthropists agree: the published PDF plus citation graph isn’t the right substrate for maintaining a representation of the evidence base in science. The pieces needed to build the alternative either already exist or are rapidly taking shape. The community is forming around exactly this problem, with concrete partnerships and shared infrastructure. A pilot should start on behavioral science RCTs because that’s the slice of empirical science most amenable to legibility, where replication ground truth is richest, and where the failure modes are best documented. What’s been missing is the galvanizing mission to assemble these pieces into something that works. That’s what I’m proposing to build. *** 1 I have a broader field map that I’ll release publicly soon. This is me, building in public! 2 If this is something that you are excited about, please reach out and talk to me. 3 More on this very soon too! 4 This tests whether, given the code and the data, you can get the same statistics as reported in the published paper. 5 This tests whether the results are the same as a paper’s given alternative analytical decisions in conducting the analysis (like outlier omission). Closely related is the idea of a “multiverse” where you come up with many different ways of answering the same underlying research question with the same data, and test whether the results hold in all those alternative methods. There’s been work on the latter as well. 6 Some of the replications, e.g. those from the SCORE project, had paired the experiments with forecasts from prediction markets. So we get to look at this for free.
WHO urges targeted action to strengthen emergency workforce capacities for shared health risks across the Asia-Pacific region
Kuala Lumpur, Malaysia, 11 June 2026 – At a time when the world is more interconnected than ever, the World Health Organization (WHO) is disseminating a comprehensive regional analysis of shared health risks across Asia and the Pacific, urging Member States to invest in enhancing workforce capacities regionwide to mitigate these risks and ensure health systems are better prepared to anticipate, tackle and withstand a range of crises and shocks. Over 100 delegates from 49 countries and areas in the Asia-Pacific region reviewed results from the regional risk analysis during the WHO-led Asia Pacific Health Security Action Framework (APSHAF) Stakeholders Meeting in Malaysia. Synthesizing insights from 21 strategic risk assessments and more than 800 multisectoral experts, the regional analysis highlighted an urgent imperative: countries must proactively invest in emergency workforce preparedness to navigate a landscape where climate hazards, disease outbreaks and geophysical events increasingly overlap. These complex converging crises create devastating, compounding impacts on health systems and communities - reiterating the need for investing in a fit-for-purpose regional health workforce and multisectoral action. Key findings: an evolving risk landscape More than half of the participating countries ranked flooding, cyclones, dengue and pandemic-potential respiratory pathogens as being of high or very high risk, while antimicrobial resistance, landslides and gastroenteritis remain widespread concerns as well. “Given that 51% - over half - of these assessed hazards are deeply interconnected, isolated health interventions are no longer sufficient,” said Dr Saia Ma’u Piukala, WHO Regional Director for the Western Pacific. “Extreme weather events like cyclones and floods directly trigger subsequent disease outbreaks, while environmental degradation - driven by sea-level rise, coastal erosion and saltwater intrusion - fuels health risks across the Pacific. Mitigating these cascading risks requires unified, cross-sectoral action linking environmental, agricultural and public health policies to build true regional resilience.” Priority recommendations for action To address critical vulnerabilities, WHO calls on Member States in the Asia-Pacific region to urgently prioritize the following strategic actions: Bridge workforce capacity gaps. Embed and standardize emergency preparedness within academia to promote regional workforce interoperability. Foster multisectoral coordination. Address ad-hoc taskforce responses and fragmented surge support through formalized contingency plans and standard operating procedures that are frequently updated and tested. Secure sustainable financing for emergencies. Overcome systemic barriers to access dedicated funding for both immediate anticipatory actions and long-term emergency preparedness initiatives. Enable emergency-ready primary health care. Embed emergency management into public health workforce readiness planning. Ensure that emergency planning is inclusive. Establish clear pathways for inclusive emergency planning that actively involves communities most at risk and other vulnerable groups. Optimize communication. Update risk communication and community engagement strategies, aligning them with seasonal risk calendars. Elevating workforce readiness to meet regional health risks While the gaps identified are significant, the Asia-Pacific region possesses solid foundational pillars, including dedicated clinicians, laboratory staff, rapid response teams, emergency medical teams and extensive community networks. However, to adequately address the compounding nature of modern crises, the various components of the emergency workforce must collaborate cohesively. True resilience depends entirely on transforming this currently fragmented workforce into a highly trained, interconnected and rapidly deployable asset. To achieve this, WHO is actively advancing regional workforce capacities through the Global Health Emergency Corps initiative. This initiative - supported by the Gates Foundation and Institute of Philanthropy - seeks to build a seamless network of emergency workforce across borders and sectors, ensuring that when a crisis strikes, a standardized, well-coordinated and trained workforce is immediately ready to respond. “Scaling up emergency workforce readiness is no longer optional,” said Dr Gina Samaan, Regional Emergency Director for WHO in the Western Pacific. “It is the definitive factor that will determine how effectively the Asia-Pacific region safeguards health and protects lives in an increasingly unpredictable future.” For more information, or to arrange media interviews, contact: Ara Johannes, WHO Western Pacific Health Emergencies Division: johannesa@who.int WHO Western Pacific Communications: wprocom@who.int