SCROLL NEWS / DISCOVERY
Search headlines
Find the stories shaping the conversation.
Results for " cheaper " 100 found 🔀 AI-powered Shuffle
Sign in to save shuffle history.
Supreme Court rulings change the economics of political party spending
A Supreme Court ruling last week allowing political parties to continue accessing cheaper TV and radio advertising may give a boost to Republicans as campaigns
GTA 6 Fans Are Already 3D Printing Items in the Game's $400 Collector's Edition
Fans of GTA 6 are already copying items in the game’s $400 collector’s edition and 3D-printing them for a far cheaper price.
Chinese scientists find a hidden atomic structure that unlocks methane
Scientists found that a tiny atomic structure that forms on nickel oxide during methane conversion is more effective than the metallic nickel long believed to drive the reaction. The discovery allowed a low-nickel catalyst to rival one containing 10 times more metal, potentially opening the door to cheaper and more efficient industrial catalysts.
Cheaper AI models may strain power grid, consulting firms say
In new reports published this week McKinsey and BCG reveal the impact of falling token prices and impact on the grid.
AirPods 5 Hands-On: Better Than ANC AirPods 4 in Every Way and Now $30 Cheaper
The AirPods 5 with wireless charging case is the best-value product that Apple announced at its 'Surprise and Shine' event.
Trump admin bars Microsoft and other tech companies from seeking permanent residency for foreign workers
he Trump administration is suspending Microsoft from seeking permanent residency for foreign workers hired through the H-1B visa program, accusing it of taking advantage of the process so that it could replace American citizens with cheaper labor from abroad.
Airlines are cutting cheaper flights amid latest jet fuel spike
Airlines are cutting more flights this year in response to high fuel prices.
Apple Upgrade: 20 Things to Know Before Leasing an iPhone, iPad, or Mac
Apple retired the iPhone Upgrade Program in favor of a new Apple Upgrade leasing option made available in partnership with Klarna. Apple Upgrade gives customers an option to get iPhones, iPads, Macs, and Apple Watches at cheaper prices than before, but there are some things worth knowing about it. We've gathered up some smaller tidbits that you should be aware of before deciding to opt in. U.S.
Killing PlayStation discs could make PS Store games cheaper, according to a former Square Enix exec
A former Square Enix business development director explains why axing PlayStation discs could make PS Store games cheaper.
Apple debuts $129 AirPods 5 with better noise cancellation and transparency mode
AirPods 5 with noise cancellation is cheaper than its predecessor.
Comstock Schools: Teachers could get raise by dropping MESSA for cheaper health plan
Michigan schools grapple with rising insurance costs
Whole Foods tries to change its ‘Whole Paycheck’ reputation with cheaper staples
Whole Foods wages war on its ‘Whole Paycheck’ past with cheaper staples
Chinese companies aren't winning just by being cheaper: Goldman Sachs
China's weak domestic demand and years of deflationary pressure have fueled concerns that its exporters could push down prices overseas.
GPU market evolves backwards as modded RTX 3080 20GB cards return as cheaper gaming alternatives
The RTX 3080 is nearly six years old, but 20GB modifications continue to appear in China. NVIDIA also continues supporting Ampere with newer DLSS models.
New Supreme Court ruling helps GOP buy cheaper TV ads for critical midterms
The Supreme Court allowed political parties to access discounted advertising rates, potentially aiding GOP candidates in the midterm elections.
3 Of The Best Pre-Built PCs That Are Cheaper Than Building One Yourself In 2026
With all the craze surrounding building your own PC, some pre-built PCs have actually become a value -- and will save you assembly time in 2026.
Is the MacBook Pro M5 still worth buying after the price hike?
It could be wise to wait, or settle for a cheaper model.
Healthcare affordability series part 8: Enrollment flexibility for two-person families can lead to premium savings
Most of the time, self-plus-one is cheaper, but there are plans where you’ll save more enrolling as self & family.
Samsung launches cheaper Galaxy Book6 with Intel Wildcat Lake, 8GB RAM from $799
The new Galaxy Book6 keeps the 14-inch 1920×1200 display but drops Core Ultra Series 3 and several ports. Its NPU is rated at up to 15 TOPS.
Why tickets for sports events are left out of DC bill combating price gouging
Buying tickets on resale marketplaces, such as Ticketmaster or StubHub, should be cheaper for shows in the D.C. in the near future.
What Made The CMF By Nothing Phones So Much Cheaper?
CMF's lower price tag is due to the brand's use of cheaper, less premium components, such as Panda glass, compared to those used by Nothing-branded handsets.
‘America wants beef:’ How Omaha Steaks is overhauling its business as prices soar
Omaha Steaks, the pioneering meat delivery company, has been around for over a century. But if it wants to survive amid record-high beef prices, the company has to become speedier, leaner and offer cheaper cuts.
America’s next big healthcare reform might already be underway
Colorado found a way to make health care cheaper. Other states are paying attention.
Deals: the Google Pixel 11 series is now on pre-order with gift cards and trade-in bonuses
The Galaxy Z8 foldables have similar offers, though, and the Galaxy S26 models are cheaper than the respective Pixel phones.
I went to Turkey for affordable healthcare; pleased with the results
A disappointing physical sent me looking for healthcare 6,000 miles from home. It was cheaper and more comprehensive.
Supreme Court rulings change the economics of political party spending
A Supreme Court ruling last week allowing political parties to continue accessing cheaper TV and radio advertising may give a boost to Republicans as campaigns
The Original Xbox Is Back, And This Time You Have To Build It Yourself
At least it's cheaper than an Xbox Series X.
5 Cheaper OLED TVs To Buy Instead Of The Samsung S95H
While the Samsung S95H is an excellent OLED TV, it can get pricey, depending on the size you pick. Here are some affordable alternatives from other brands.
Cheaper health insurance changed their lives. Then it went away.
Part 1 of an occasional series on how people are being affected by the expiration of Affordable Care Act subsidies. EDWARDS, Miss. — Every now and then a freight train rumbles through, creaking and groaning on the rusted tracks that run along the edge of the largely deserted downtown. On some days, that train is […]
AI searched 100 million possibilities and found a cheaper way to 3D-print a NASA rocket alloy
Researchers used AI to search through more than 100 million possible settings for 3D-printing a high-performance NASA alloy. After only 40 experiments, the system identified six successful configurations, including one that worked at a record-low 500 watts. That could allow GRCop-42, currently difficult and expensive to print, to be made with much more widely available equipment.
‘He Could Be a Cheaper Option’ — Sports Media Expert Theorizes ESPN May Be Planning to Replace Troy Aikman With 3-Time Pro Bowler
The longest-standing broadcast partnership in NFL history may be approaching a crossroads, and one media analyst has floated three-time Pro Bowler Greg Olsen as a possible option for ESPN if Troy Aikman departs. When Troy Aikman, the on-air partner of Joe Buck since 2002, revealed on the “Sports Media with Richard Deitsch” podcast that he has not signed a contract extension with ESPN, rumors swirled.
10+ Huge SEGA Games Are Cheaper Than Ever In New Xbox Publisher Sale
A particularly fantastic new set of offers!
Fall's Video Game Release Slate Is Intimidatingly Good And Completely Doomed
Seemingly every game ever is dropping, but how many will survive?
Op-ed: Utility bills are rising for the sake of scoring political points
When the cheaper option is also the cleaner one and some leaders still fight for the expensive one, affordability clearly isn't their priority.
New soil test could reveal how soil microbes help feed plants
Scientists have streamlined a method for measuring DNA-bound phosphorus in soil, making it cheaper and easier while preserving its accuracy. The technique could reveal how soil microbes help recycle phosphorus and ultimately support more sustainable fertilizer and nutrient management.
Gen Z women want politicians who can make the ‘American dream’ feel attainable
An organizer with the Ohio Student Association said students bring up the cost of gas, weighing if it’s cheaper to commute or live on campus. They also talk about whether to get an unpaid internship or work a job out of their field of study to save money.
Apple and Schneider’s new smart thermostat could help cut your electric bill
Apple and Schneider's smart thermostat uses your iPhone to adjust temperatures and can shift energy use to cheaper, cleaner hours.
Scientists just found a new way to make hydrogen from water
A new light-activated material can efficiently produce hydrogen from water without requiring an additional expensive metal catalyst. Researchers say its unusual sulfur-based chemistry could point toward cheaper and more practical ways to turn sunlight into clean fuel.
The iPad Mini is slightly cheaper again during Prime Day
Save $100 during Amazon’s October sale.
State Fair of Texas brings lower prices, new attractions, more value options
From cheaper rides and games to over 200 food and drink options priced at $10 or less, the fair is putting a bigger focus on affordability.
The iPhone Duo enters China’s crowded foldable market — and faces a price test
Chinese netizens focused on the price of Apple’s iPhone Duo, as cheaper Xiaomi and Huawei foldables raise value comparisons and muddle the foldable's reception.
Dell’s new 14S laptop exists to challenge MacBook Neo and, somehow, might be even cheaper
Apple’s MacBook Neo has made waves in the laptop world because it’s a remarkably well-built and reliable laptop at a...
With new Macs, Apple aims to take on Microsoft, Nvidia in a rush to lower AI costs
SAN FRANCISCO, Sept 22 (Reuters) - When Apple's new desktop computers start shipping Tuesday, the company's executives will make an unusual pitch to corporate buyers: They are cheaper than renting data centers. Apple's upgraded Mac Minis and Mac Studios which can cost nearly $20,000 and handle AI tasks locally will go head-to-head against new desktop machines that Nvidia and PC makers are rolling out, expected to be the focus of a Microsoft Windows event in San Francisco next month.
The Budget Android Tablet Brand That Can Compete With Samsung Galaxy Tabs
The Teclast T65 is cheaper than an Apple iPad or Samsung Galaxy Tab, but the reason is partly its chips are slower. That affects performance.
GTA 6's $100 PS5 Ultimate Edition Is Getting More Pre-Orders Than the Cheaper Version
But that's to be expected
OpenAI cuts prices on smaller models as businesses scrutinize AI spend
OpenAI slashed prices of its low- and mid-tier AI models on Thursday, a move that may intensify competition in the industry as U.S. companies battle cheaper Chinese rivals for customers increasingly wary of the technology's ballooning costs.
Rumour: PS6 Hardware Complete, Double the Power of PS5 with Handheld Model
Less powerful than new Xbox, but potentially cheaper
Jonathan Kuminga’s Agent Defends Reason For Taking Shorter and Cheaper Deal With Timberwolves
On Wednesday, ESPN’s Shams Charania reported that Jonathan Kuminga had agreed to join the Minnesota Timberwolves on a two-year, $12. 4 million deal.
Gleyber Torres’ Tigers Future Just Got Complicated
Gleyber Torres grounded into a game-ending double play, and Detroit already has a cheaper option waiting. #Tigers
Sony finally has a cheaper OLED to compete with midrange Samsung and LG TVs
The Bravia 6 is Sony’s first OLED TV with four HDMI 2.1 ports.
I went to Turkey for affordable healthcare; pleased with the results
A disappointing physical sent me looking for healthcare 6,000 miles from home. It was cheaper and more comprehensive.
Judge won’t halt Atrium’s cheaper copays for State Health Plan members, for now
After the ruling, the State Health Plan warned premiums could more than double in 2028.
Our favorite headphones and earbuds are cheaper during October Prime Day
Spend a lot or a little.
Deals: Galaxy Z Fold7 vs. Galaxy Z Fold8 and Z Fold8 Ultra, plus Galaxy S26 FE alternatives
The Z Fold7 is a lot cheaper than its replacement - but you will be missing out on some great improvements and the Z Fold8 Ultra is discounted as well.
September Monthly Review: Health MCP Guy?
For all the time I spend talking about APIs and data exchange, this newsletter has had embarrassingly little interoperability of its own. It’s old-school SaaS, baby - UI or bust. Not all of that is my fault! Substack, for all its popularity, is (as mentioned before) really frustrating as my system of record. The primary user it optimizes towards is the reader, whose attention feeds the app, the Notes feed, and the recommendation network that Substack is actually selling. Health API Guy is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Subscribe They’re falling into the same trap that Medium once sprinted into, albeit at a much more glacial speed. Younger generations may not remember (or never knew) it, but Medium once lured publications like The Ringer and Signal v. Noise onto its platform, only to lose them as it rebuilt itself around a paywall and an algorithmic feed. Substack’s original pitch circa 2020 was the antidote (you own your list), but every new feed feature chips away at it. I get it: writers are generic supply in the eyes of the Substack product team. Generic supply follows demand, so the consistent pressure on the PMs is to generate demand, which leads to every product cycle going toward discovery (Notes, recommendations, the app, live video). The end result lackluster author tools (laggy and broken analytics are truly crippling) all despite the premium price point. The archive is where it crosses from authorial gripes to actual reader impact, though. Across Substack and LinkedIn, I’ve published 1,218 posts and nearly 900,000 words since 2018. Substack’s native search can see only the 464 posts that live on Substack, and even there it’s woefully inadequate at surfacing relevant content (even for the guy who wrote it)! Readers have the same problem, writ large. Ask the Archive Fixing search (for both my readers and my own use) turned out to be extremely non-trivial, unfortunately: There’s no public API for readers to access content RSS exists for new content notification, but it only carries the 20 most recent posts and cuts paid ones off at the preview The only Substack MCP is an admin tool for analytics The export exists for authors, but it’s painfully awkward and slow. So I built something to work around that: the Health API Guy MCP. It’s a private MCP server holding the full text of everything I’ve written across Substack and LinkedIn. Add it to Claude or ChatGPT as a connector and you can ask things like: It pairs keyword search (for acronyms and case names, of which we have many) with semantic search. Every post is also labeled by topic, form, and court case, so you can pull a lawsuit’s (or multiple cases’) entire docket across Substack and LinkedIn in one query: Hail, Healthapiguyhydra This is also a bit of me eating my own cooking. I’ve spent the year arguing that headless healthcare is arriving and that developers want primitives over abstractions. Last week in Hail, Benihydra, I watched Salesforce plug its platform into Claude and bet that the records underneath outlive whatever interface sits on top: Salesforce must believe that the Hydra’s immortal head is the records, permissions, and business logic (in that they will survive whatever happens to the interface). It felt a bit hypocritical to keep my own archive behind a Substack search bar. The archive is the body. Substack is one head, and now whatever tool you might use is another. Will the EHRs follow my lead? Probably not imminently, but one can hope! Why Founding Members Only The MCP is available to Founding Member subscribers. A few reasons: It costs money and time to run. Hosting, the database, Clerk usage and AI compute for search and labeling are significant operational costs (that increase with usage). Substack gives me no way to automate access. There’s no API to check whether someone is a paid subscriber, so every user is manual operations for now. It’s a beta. We’ll need to improve things and parts will break. I’d rather they break for a small group of dedicated readers who will tell me about it. The Founding tier deserved a real benefit. Until now it was mostly a generous tip. Now it comes with something. If you’re already a Founding Member, reply to this email with the address you’d like provisioned. If you’re not and want in, upgrade here. Larger teams (such as companies) can get access too: group subscriptions at the Founding tier come with the group discount. The admin for the subscription can reply with the list of addresses to provision. Get 20% off a group subscription Pricing Founding membership is now a flat $200 a year, replacing the old suggested $150 that readers could adjust. Existing Founding Members will have access grandfathered in at their previous rate. Month in Review Articles Published Make Networks Dumb Again (Sep 11): AI is making old analog networks cheaper to use and flooding them with machine traffic in the process. A plea for agent-native rails with the intelligence at the endpoints, and for Washington to drag them to ubiquity. Interoperability, Unbuttoned (Sep 29): Interoperability is a series of workflows in a trenchcoat, and this post finally takes the trenchcoat off. A color-coded map shows which provider workflows went digital, which are fragmented, and which still run on fax. Video Content Antitrust’s System of Record Problem (Sep 3): The FTC’s investigation into Epic prompted a ninety-minute webinar, now posted with chapters and a smoothed transcript. We get into why market definition, missing B2B share data, and the SSNIP test leave antitrust poorly suited to systems of record, plus why information blocking may be doing more than any pending lawsuit. The Information Exchange: Cerner Milking Edition (Sep 18): This was so fucking fun. Ryan Tucker, Ryan Brickner, Brad Thorson, and I try an Around the Horn format, complete with points, penalties, and a mute button. We debate whether Oracle Health is dead, who would actually buy Cerner (IBM enters the chat), p(Doom), and why the front desk still hands you a clipboard. Regulatory Ask Not What Your Network Can Do (Sep 23): CDC’s RFI on public health data intermediaries lists use cases already served, unevenly, by eCR, lab reporting, IZ Gateway, and syndromic surveillance. The real opportunity sits in one question about AI: a network that can handle the next emergency without another interface project. Court cases Particle v. Epic: The Naughty Schoolteacher (Sep 4): After a summer-long vigil, the answer on Epic’s early summary judgment bid is two paragraphs and a remedial assignment. Particle advances to the next grade, and market definition gets another year of discovery. The Epic Discovery Files (Sep 8): Texas’s motions to compel arrive with a 251-page appendix of requests, objections, and meet-and-confer letters with Cravath. Inside: Epic using generative AI for document review, denied interoperability requests in scope, and Noerr-Pennington raised against lobbying discovery. Epic v. Health Gorilla: “Working as Designed” (Sep 9): Epic wants nothing to do with the data breach MDL, and explaining why puts fresh discovery on the public docket. The spiciest allegation: Health Gorilla told GuardDog to scrub its law firm business from its website instead of cutting off access. Veeva v. Epic: Legal Bypass Surgery (Sep 15): Veeva asks the Wisconsin Supreme Court to skip the Court of Appeals, while law professors and three policy groups line up as amici. Meanwhile, Epic’s lawyers apologize for another round of citation errors, including a quote that never existed. Texas v. Epic: Behind the Black Bars (Sep 17): The court portal served up Epic’s unredacted filings, revealing a $5 million document review budget and the AI workflow behind it. Also inside: the 200-plus company names Texas wants searched, and one very expensive Zoom invitation. OpenEvidence v. Doximity: Consequences Will Follow (Sep 25): OpenEvidence tells the court it will not comply with a discovery order, and its own lawyers at Quinn Emanuel say they won’t defend the choice. Add years of Slack reported unrecoverable, and Rule 37 sanctions come into view. Health Tech Keeps Choosing Violence (Sep 28): Doximity moves to end OpenEvidence’s case, Epic asks Texas to show its work, and a sleepy Audacious Inquiry patent fight sprouts antitrust and information blocking counterclaims. The main event is a long-sealed Cognizant v. Infosys ruling that hints where system of record antitrust claims can survive. EHRs One Verb for Oracle Health (Sep 14): Oracle’s earnings call quantifies every data center to the decimal point, while Oracle Health gets a single verb: “accelerate.” Panning the few healthcare mentions turns up an agentic care management system, a research-to-care ledger play, and the question of whether healthcare gets Larry’s patience. Epic’s Forretress Under Siege (Sep 30): Judy Faulkner tells a panel Epic paused hundreds of projects to harden its software and calls it a “shame,” while a spokesperson insists the roadmap hasn’t changed. Project Glasswing, a Pokémon-branded QAN sprint, and a trove of Jodel grievances reconcile the two Epics and preview the Jevons paradox coming for all software. Industry Analysis Scanning for a Network (Sep 1): Scan.com raises $220M to build the imaging network labs never got, speedrunning Zocdoc’s consumer-to-B2B pivot. Its wedge to network density may be the most American one available: personal injury litigation. The Front Desk Wants the Damn Whole Building (Sep 16): Hello Patient buys Converse Health, pushing its AI front desk into referrals, chart work, and prior auth. The copilot categories keep blurring, and your distribution partner is still your final boss. Cross-industry Comparisons A Different Patchwork Quilt (Sep 10): Healthcare loves to envy open banking, but the US version has no mandate, a toll booth at Chase, and data quality problems of its own. The grass isn’t greener; it’s another American patchwork quilt, and perhaps a worse one. Hail, Benihydra (Sep 21): Salesforce brings its platform into Claude, Slack, and its own agent, inviting the interfaces that might replace it. Cut off one head and more grow back, so long as the records underneath stay put. Living on A Trade Secret Prayer (Sep 22): Six credit unions are suing Fiserv, the Epic of core banking, by claiming their own member records as trade secrets. Banking has no Cures Act, so its lawyers reach for legal alchemy that healthcare has already tried. Other News None this month External Media Nabla Accelerate NYC: I kicked off fall conference season at Nabla’s Accelerate event in New York, where Delphine Groll opened by asking the room to picture the company it wants to become and the one it doesn’t. I liked her framework of bureaucracy, fear, and ego as the blockers. I also love vendor conferences/events in general - different vibe than industry biggies. Healthcare 101 with Nikhil Krishnan: I briefly guest lectured Out-Of-Pocket’s Healthcare 101 class. Appreciated Nikhil’s takeaways cover whether AI doctors can query HIEs, why healthcare’s data-liberation rules beat most industries’, and the hi-res graphic is in the comments. HCN: Epic Probed by the FTC: I joined the HCN guys to talk through the FTC’s investigation into Epic. Cedric’s new hairdo was an unexpected plot twist. The episode also covers MFN drug pricing, healthcare M&A and Eli Lilly’s acquisition streak. Fall Conferences: At eHealth Exchange’s Annual Meeting in Austin on Oct 27, I’ll moderate a panel on the tensions between AI, HIEs, and timely patient access with Jean Ross (Primary Record), Therasa Bell (Kno2), Byron Crowe (Doctronic), and Michael Marchant (Freenome). Here’s the rest of my schedule if you want to meet up: CommonWell (Redwood Shores, CA): Oct 13-14 Will be participating in a fun Jeopardy event Open@Epic (Madison, WI): Oct 20-23 Will be enjoying Verona, Willy Street, and other local adventures eHealth Exchange and Sequoia Project (Austin, TX): Oct 27-29 Yeehaw HLTH (Vegas): Nov 15-18 Groundhog Day, Venetian Edition RSNA (Chicago): Nov 30-Dec 2 See me at “Governing AI in Breast Imaging: What Health Systems Are Actually Doing” Digital Health Counsel 2026 AI Summit (Seattle): December 2-3 Will likely be speaking to a bunch of lawyers about information blocking Posts I Liked ChatGPT’s Epic integration: game changer or nothing burger?: Joshua Liu, MD offers five thoughts on OpenAI’s Epic launch, starting from the point that it rides the same open APIs available to anyone. He argues read-only access leaves ordering to Art and the ambient startups, and asks why HCA is piloting it alongside Commure. He’s sorta a must-follow at this point - consistently excellent, thought-provoking takes. Epic, Lilly, and the Discovery pass: Seth Chaney reads Epic’s first major Discovery deal (with Eli Lilly), as well as its revised developer terms on commercial displays, as the pipe finally opening between EHRs and pharma. Life sciences and health continue to collide. Scott Rossignol on EHR payer platforms and ePA fees: Scott argues the top ambulatory EHR vendors are turning the prior auth mandate into a per-member-per-year toll on payers, for an API whose operating cost has nothing to do with covered lives. His fix is two lines of regulatory text: put the CRD/DTR/PAS certification criteria in the Base EHR definition, and bring payer-facing ePA connectivity under § 170.404’s cost-based fee rules. It’s a good solution. Health API Guy is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Subscribe
The reason the Mac mini is cheaper than a MacBook
Is it better to buy a Mac mini or a MacBook? If you already have a display, mouse and keyboard, the mini makes sense. If you don't, a MacBook might be smarter.
5 Streaming Devices Cheaper Than The Roku Ultra (After The Price Hike)
Streaming devices that serve as worthwhile alternatives to the Roku Ultra but don't cost as much include TiVo Stream 4K Onn 4K Plus, and Roku Stream Stick 4K.
Opinion: Chicago is losing affordable housing faster than it can build it
Chicago can’t solve its housing affordability crisis through new construction alone. Preserving existing affordable units is cheaper and faster.
New nanoparticles make hidden chemical differences light up
Scientists have developed ultra-bright nanoparticles that can detect tiny amounts of chemicals and distinguish between molecules that look almost identical. The technology could one day provide a cheaper way to catch dangerous drug impurities or trace pollutants using simple low-cost lasers.
United's long-haul business class fares just got cheaper
United Airlines just made some lie-flat business class seats cheaper on long-haul flights, but there's a catch. What to know about the new Polaris base fares before buying.
Samsung Galaxy Z Fold8 Ultra Is Now Cheaper Than the Standard Z Fold8 After Hitting a Record Low
For a limited time, you can save as much as $300 on the Samsung Galaxy Z Fold8 Ultra at Amazon.
5 Cheaper Android Tablets To Buy Instead Of The Samsung Galaxy Tab S12 Ultra
If you're in the market for a new tablet and have been eyeing Samsung's Galaxy Tab S12 Ultra, you might consider these cheaper alternatives.
Lenovo's first Googlebook will cost less than you (and Google) thought
The Lenovo Googlebook 15 is one of five Googlebooks as part of the first wave, and it turns out this device is cheaper than first claimed.
Trump wants cheaper beef. Republicans in key races warn his plan goes too far
Senate Republicans, including Tom Cotton and Tim Sheehy, push back on Trump's plan to import 300,000 metric tons of foreign ground beef to lower prices.
Drugs for autoimmune diseases are pricey. This bill could switch patients to a cheaper one
Nearly 25,000 Californians could get switched to a cheaper drug next year — unless their doctor objects.
Supreme Court rulings change the economics of political party spending
A Supreme Court ruling last week allowing political parties to continue accessing cheaper TV and radio advertising may give a boost to Republicans as campaigns
Streaming’s Biggest Show Returns…and a Buzzy Genre Struggles
(Welcome to my weekly streaming ratings report, the single best guide to what’s popular in streaming TV and what isn’t. I’m the Entertainment Strategy Guy, a former streaming executive who now analyzes business strategy in the entertainment industry. If you were forwarded this email, please subscribe to get these insights each week.) Obviously, the big news of the week (if not month) is Netflix letting go of executive Bela Bajaria, their chief content officer. I’m going to cover this in my next “Most Important Story of the Week”, but I also have a data angle that I might publish before then. (Too much news this month, am I right?) But first, I have a bonus Streaming Ratings Report issue for you. Today’s article is a bit of a grab bag of topics I’ve wanted to write about, but couldn’t squeeze into the (already very long) Streaming Ratings Report issues. The NFL drives huge viewership numbers, so I’ve got a look at that. A whole bunch of TV shows returned to the traditional broadcast channels, and I have thoughts. And I have a quick look at adult animated shows, since a couple flopped earlier this month. Plus, the Emmys aired earlier this month, and we got some fun data bout how popular those nominated shows are and why I’m not optimistic about the show’s future. That’s a lot, so let’s dive right in. Subscribe Sports - Football Remains the King of Streaming Over the last two Streaming Ratings Reports, I sorta ignored the biggest “show” each week. (I was tempted to say “piece of content” but I know that word bugs folks.) In my database, I call these “specials”, and they sure are special. I speak of the king of American TV: football!!! Streaming football is back and, unlike some other sports, it routinely dominates all viewership. Probably the best way to show this is to update one of my favorite graphics, the “40 Million Hour Club”1 since football games have joined that club each week they’ve premiered: (Yep, that’s the data through 28-Sep, the first four weeks of the season. As a reminder, I take the average minute audience provided by Nielsen, factor out a percentage of linear TV viewers in home markets, and multiply the run time by the three-hour runtime.) Again, the numbers are huge. Let’s start with Netflix. They grabbed the rights to the bonus opening week NFL Kickoff game (which was on YouTube last year) and it put up 18.5 million viewers on average. That nets out to 47.4 million hours of estimated streaming hours. That’s good! Though, notably, it’s well below Christmas Day games from previous years. Compared to other streamers/broadcasters, that’s a bit of a mixed result. On one hand, Netflix’s 18.5 million average is about what YouTube claimed last year, though Nielsen used a bespoke methodology at that time, so I trust this viewership figure more. So that’s good. (The game also shifted from a Friday to a Thursday and that could have helped the viewership as well.) And Netflix continues to best Prime Video in total viewership—as you’ll see next—though Netflix has had much higher-profile (opening week and Christmas Day) games. NBC/Peacock’s opening game the night before averaged 25.1 million average viewers, which is just much bigger than Netflix (or Prime Video). That’s the big caveat to all streaming viewership records: even in success, they can’t compete with traditional linear TV. Previously, I called this my “hierarchy of viewership”, and the data keeps supporting that thesis. The downside for NBC is that this game’s viewership was down compared to previous years, though it may be a counting issue. Here’s a great chart from Sports Media Watch: Indeed, looking at all the opening week games across all windows, Netflix still lags the other primetime windows, but does well compared to the afternoon games: I saw a few folks make the case that streamers still lag traditional broadcasters in reach, and I’d add they really lag compared to “broadcast plus streaming” options like NBC/Peacock, ESPN/Disney+ and CBS/Paramount+. The best data I saw for this came from Samba TV, who charted reach through the 2025 season, and, wow. (Samba TV’s numbers are historically lower than Nielsen’s, so don’t compare their household figures to Nielsen’s hours, but use them as a proxy for the relative reach of the streamers.) The good news is Prime Video is NOT having a down year. Their opening Thursday Night Football game reached 18.6 million viewers (about what Netflix got), but that’s a high for Prime Video on Thursdays and the biggest Thursday night game since 2015. The TNF games the weeks of 21-Sep and 28-Sep continued this streak, getting 43.1 (16.8 million average viewers) and 42.5 total hours (16.6 million average viewers), which were both up from last year. Still, that continues the (typical) season slide: (For those new-to-streaming-football-numbers folks, the viewership declines for Thursday Night Football as the year goes along.) Overall, it appears as if Prime Video is now part of the “football habit”, be it for customers at home or out at places like bars and restaurants. And its viewership is slowly catching up to broadcast. One last chart shows this. Last season, Nielsen started adding out-of-home viewership to the charts, so I try not to do season-to-season comparisons as I did for the first few years. And while Nielsen added Big Data viewing this year, I’m going to compare the last two seasons. Here’s that chart, which shows the season-over-season gains: When it comes to streaming football, we have to keep a few seemingly contradictory thoughts in our head… First, streaming doesn’t come close to matching broadcast’s reach and, thus, viewership. Second, even with that gap, football is still the biggest “thing” on streaming. It’s very, very popular. Third, while some methodology changes have likely helped Prime Video’s numbers, their viewership does appear to be growing. Lastly, yeah, the NFL started the season off poorly, with week one dragging down the numbers in year-over-year comparisons. However—and again, methodology changes may have helped this—the viewership grew compared to previous seasons over the next couple of weeks. So football is still king. On streaming and off. Television - Adult Animation Continues to NOT Make the Streaming Charts In the last two reports, two adult-animated shows whiffed: Netflix’s animated comedy, Dang! Paramount+’s Sega-video-game-inspired fantasy/comedy, Golden Axe. Neither made the charts, but that’s usually the case for animated shows. I mean, here’s literally every animated show that’s made the Nielsen Original charts since the start of 2025: And one of those shows is G_abby’s Dollhouse_! A kids show! The problem with charts like this is that they don’t give you a sense of how many shows miss the charts. Netflix had at least eleven animated flops, including comedies like Big Mouth (a show everyone assumed was popular in the pre-Streaming Ratings Era), Long Story Short, Haunted Hotel, Mating Season, Strip Law, and Alley Cats, and action shows like Blood of Zeus, Castlevania: Nocturne, Devil May Cry, Splinter Cell: Deathwatch, and Tomb Raider: The Legend of Lara Croft. Prime Video had at least six animated misses, two in the same universe (Hazbin Hotel, Helluva Boss), three more comedies (#1 Happy Family USA, The Second Best Hospital in the Galaxy and Kevin) and an action show (The Mighty Nein). Disney+ had MCU and Star Wars animated misses (Eyes of Wakanda, Marvel Zombies, Star Wars: Visions, Star Wars: Maul – Shadow Lord, and Star Wars: Tales of the Underworld) but I can’t tell how many of those shows were geared towards adults. HBO Max had Harley Quinn and Hulu had Solar Opposites. None of these shows made the charts! That’s a lot of misses! If I remade the chart with all those misses, here’s how it would look: Listen, I know the counter arguments. First, it’s cheaper to produce animated shows than live action. Fair enough, but with viewership this low, even that cheaper price point starts to not matter. Second, just look at the acquired charts. Hit animated comedies can generate viewership for years! Shows like The Simpsons, South Park, Family Guy, and Bob’s Burgers do really, really well. Fair enough, but streaming has failed to recreate the dynamic that Fox has mastered so well. That’s a problem for streaming still. Television - Emmys! The annual Emmys broadcast was a few weeks ago, and I’ve got data on how the nominees performed on streaming, plus hot takes on this year’s viewership (and the big news about the broadcast’s, er, streamcast’s future). We’re just getting started with this issue, but the rest is for paid subscribers of the Entertainment Strategy Guy, so if you’d like to find out… The top Emmy-nominated shows… ...why I’m worried about the future of the Emmys telecast... ...the pros of broadcast television ...and the cons... ...please subscribe! We can only keep doing this great work with your support. Coming Soon! Next week, we’ve got a double issue (since the weeks of 21-Sep and 28-Sep were so light), but then I’ll be all caught up after a deluge of streaming ratings data from Nielsen in September. In the meantime, I’m going to try to take a look at Netflix’s stats to explain why Bela Bajaria is out. But even light weeks give me plenty to talk about. The Great British Baking Show returns to Netflix and Apple TV has a buzzy new star-studded comedy. And Netflix has a controversial-but-appears-to-be-popular new game show. Plus big films like Toy Story 5, Backrooms and The Breadwinner head to streaming. And is East of Eden a hit? After that, The Kardashians return to Hulu, Prime Video has a reboot of Carrie, and Netflix has a horror show, Below. Disclosure Day heads to Peacock, The Housemaid heads to HBO Max_,_ a new Ben Affleck film, Animals, lands on Netflix, and Apple TV has a new John Cena action film based on, checks notes, Hasbro’s Matchbox toys. Okay. Longterm, some interesting TV production news. HBO Max ordered How to Survive Without Me, starring Kaley Cuoco, Joshua Jackson, Ray Romano and Rita Wilson, for fifteen episodes! I love that some streamers (or, at least, HBO Max) are finally ordering TV shows with more than eight episodes. In less wise production news, Landman does not appear to be coming back this year, since it just started production. Frankly, I don’t love this (get your shows out on time) but Landman is so popular, I doubt it matters.
Supreme Court rulings change the economics of political party spending
A Supreme Court ruling last week allowing political parties to continue accessing cheaper TV and radio advertising may give a boost to Republicans as campaigns
5 Budget Android Phones You Should Consider Instead Of The Google Pixel 11
Considering a Google Pixel 11 smartphone, but feeling put off by the $900 price tag? Why not consider these five much cheaper Android phone alternatives.
A popular GLP-1 provider just made its weight loss medications cheaper
Mochi Health has lower starting prices for compounded semaglutide and tirzepatide, plus a new lab testing service designed to help patients monitor their health.
Trump's Belarus potash plan draws warning over Russia route
Belarus says this year's potash output is already contracted as Trump seeks a cheaper supply deal with the US.
Moving abroad helped me feel healthier; eat better for cheaper
Moving to the Caribbean and leaving the US didn't cure my health issues. It changed my relationship with wellness and made produce more accessible.
The Xteink X4 Pro is a great, tiny e-reader, and it’s cheaper than ever
It’s the best Xteink e-reader yet.
Supreme Court rulings change the economics of political party spending
A Supreme Court ruling last week allowing political parties to continue accessing cheaper TV and radio advertising may give a boost to Republicans as campaigns
Has Oracle Become A Weaker Business, Or Just A Cheaper Stock?
Oracle (ORCL) made almost a quarter more operating profit in its last fiscal year. Yet when trading closed on September 10, the stock sat around 53% under the peak it reached a year before.
It's literally never been cheaper to buy this Wahoo indoor trainer on Amazon thanks to a whopping Prime Day discount
Shorter days, colder weather and more time indoors are coming so now is the time to upgrade your indoor trainer
SCOTUS grants Republican appeal on TV ad rates before midterms
The order clears a legal roadblock to political parties getting cheaper TV ad time ahead of the midterm elections
Campaign cash from utility leaders prompts fight in governor's race
The candidates for Michigan governor vow cheaper energy bills and attack one another over utility leader and PAC campaign contributions.
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.
Sundar Pichai: Google’s Frontier Imperative, Compute Scarcity Dilemmas, Future AI Products and Businesses
Alphabet CEO Sundar Pichai has little patience for doubts about Google’s willingness or capacity to compete at the frontier of AI. His company’s latest Gemini 4 Argon model — long-promised and recently delivered — beats OpenAI and Anthropic’s latest public models on coding, math, science, and knowledge-work benchmarks and has placed Google’s AI back among the world’s best. Amid chatter that Google was fine with good-enough lightweight ‘Flash’ models, Pichai pushed back hard against any suggestion that company is limiting its research ambitions. “We won’t do it any other way,” he told me in a Big Technology Podcast interview recorded this week at Google’s Mountain View headquarters. Pursuing cutting-edge research is costly for Google, requiring compute and chips that otherwise could be sold at a high markup by its cloud business. But Pichai said allocating resources to frontier AI development is “ground zero” for Google. “Our cloud business is at any given moment in time forgoing short-term revenues because we are making compute trade-offs to support frontier AI development,” Pichai told me. In our conversation, we discussed Argon (which Pichai says processed three times as many tokens as Gemini 3 in the 72 hours after launch), AI’s role in developing better models, the AI interface of the future, competition with OpenAI, Google’s partnerships with Anthropic and Apple, data centers in space, and the nature of his optimism. You can read the full conversation, lightly edited for length and clarity, below, or listen to it on Apple Podcasts, Spotify, YouTube, or your podcast app of choice. Alex Kantrowitz: Google is back at the frontier with Gemini 4 Argon, which beats OpenAI’s and Anthropic’s public models on benchmarks spanning coding, math, science, and knowledge work. The conventional wisdom was that Google had given up on the frontier to focus on fast, good-enough models for its products. That was wrong? Sundar Pichai: Very wrong. Look, we are so committed to being at the frontier. It’s in our DNA. We won’t do it any other way. We’ve been thinking about the AI frontier for well over a decade now. The frontier is very dynamic. It’s very competitive. It’s very exciting. I couldn’t be more pleased with how Argon has come together. We took our time with it. We made sure we tested it safely and deployed it carefully. We took a bit longer than our other models, given its capabilities. Across the board, be it for agentic work, knowledge work, coding, creative tasks, Argon is one of the really good frontier models. The first 72 hours since we launched, it’s almost up to 25 trillion tokens used, which is about three times what Gemini 3.0 had in the first 72 hours. And we are capacity constrained as we are serving it. Excited, but there are always areas….. You know, the frontier is very dynamic, and you have to keep evolving the models. There are areas which we can see we need to improve as well. We’ll be at work on all of that as well. Which areas? On any given model, Argon would be across the board better at everything if you just look even two months back. But the space is so evolving. Even on coding, for example, there are pockets where we know we need to do better, and there’s clear feedback. We’ll make it better. The argument for standard models is that they’re already good enough to build great products, with the frontier valuable only for certain tasks. Meta built Muse with good models rather than the best, and Google could put standard models into Maps, Gmail, and Search. Why isn’t that sufficient? It’s definitely not how I have thought about it at all. We are deeply committed to driving the state of the art and actually pushing where this technology can go, all the way from transformers. That’s a journey we’ve been on. You will see us continue to do that. When you’re at the frontier, you get an earlier look into the emerging capabilities, earlier than others, which helps you better prepare for the future. We are fortunate to have a wide set of opportunities across consumer and enterprise. Some of our cloud customers, for sure, want the model at the frontier. The frontier models will be better at developing the next set of models in an era of recursive self-improvement. I would argue it matters there. Internally, we have definitely seen a steep point of inflection by deploying Argon to our developers. It definitely tackles problems which our previous models could not do. Being at the frontier puts you at the innovation frontier and helps you drive that as well. There is room for models of all sizes and capabilities. That’s why we did Flash, Flash-Lite, etc. I agree with you that you can develop very good product experiences with not necessarily the most frontier model, because they tend to be slower and more expensive too. But we care about the Pareto frontier, and we’re committed to doing all of that. You said developing frontier models gives you a look at what’s coming. What do you see on the horizon? What makes this moment so exciting is you build something which you think is a culmination of months of work, and you put it out. You already see clear headroom ahead. We’ve all been working on these agentic, long-horizon capabilities, but clearly the frontier is going to have multiple of these agents working together to do things. That’s an example of an exciting paradigm underway. It’s also clear to me that users, both in consumer and enterprise, will have access to some multi-agent framework for them, working all the time doing things. It feels like a moment of infinite possibilities. Sergey Brin said one of Argon’s core innovations was relying more on the model to help develop the model. Yet incidents involving OpenAI’s bots breaking containment and Google’s bots accessing a site they shouldn’t have suggest these systems still aren’t perfectly aligned. Isn’t there reasonable cause for concern about accelerating toward recursive self-improvement? There are three parts to your question. Let’s first talk about RSI itself. You’re right, Sergey mentioned RSI being a critical contributor as we are developing Argon. But to be very clear, today, in a continuum, we are pretty far from the end state of RSI or something like that. The right way to describe it is you have RSI loops along with AI researchers in various aspects of model development. You almost have agents helping AI researchers in some automated way work together to build these models. It’s good to at least understand where we are today, and it’s super exciting. I expect that continuum to keep making progress as we go into next year. The second question is around safety. Security comes from defense in depth. You’re talking about alignment, obviously. We all are doing a lot of work to make sure the models are as aligned as possible. But it’s an active area of research, and there are clear gaps. That’s not the only thing we rely on. You have the work you do on alignment, but beyond that, we put a lot of effort into real-time monitoring as these models are working. Making sure you have good telemetry into what’s happening so that you can intervene in real time if you see unexpected behaviors. How you do that, the architecture with which you do that, how you make sure it’s all sandboxed and isolated in a way that it can continue doing its job during real-time monitoring — these are all areas of investment. Obviously, when we deploy these models, giving customers, enterprises, control, both with real-time monitoring, but their ability to decide where they deploy, what type of data which models ever have access to, etc. All that work is very, very important, and we are doing a lot of that. As an industry, we need to constantly up our game and take the time to do all this stuff, including dedicating the compute required to make sure you can safely deploy these models. The third part is: Is it all enough? This is why there are a lot of conversations in the industry, and I think it’s reasonable to be worried. The right way to approach something you worry about is early on thinking about what more we can do. As an industry, you’re definitely seeing increased levels of work on alignment, testing for safety, etc., as we deploy these models. The next step would be developing better standards so there are cross-industry standards of how we do these things, including the ability to evaluate using independent evaluators. If you look at areas like financial controls, we have accounting standards, we have independent auditors. Those are all the next natural steps for us as an industry to work toward, and regulation will end up playing a role as we go through all of that as well. Google pursues both language models and world models, while some other AI companies like OpenAI have focused more heavily on LLMs as the path to AGI. Why is world-model research important to you? As humans, you obviously perceive the world around you, and you’re taking all those inputs. Making sure these models are able to understand the laws of physics, simulate the world in high fidelity. All that, I think, are key enabling blocks for intelligence. We are working across all possible applications of this. Take an area like robotics. It’s not an accident: If you look at embodied reasoning benchmarks, Argon is state of the art. That all comes from our experience building world models. Inherently, I would expect there will be architectures which will converge all of this in the future. But for us, we view this as, today, this jagged frontier that is intelligence. We want to be at the frontier of all the possible approaches, and that will help us both drive better research, better insights, and develop a much better AGI model, is how we think about it. Some people argue Google should focus on selling its computing resources rather than using them for AI research. Cloud wants to sell compute at a high markup, while researchers need a very big computer to build the best models. How do you balance those demands? Ground zero for us is making sure we are investing in the compute that’s needed to be at the AI frontier. In 2015, I set the company to be AI-first. AI is that key horizontal ingredient across everything we do, all the way from Search to Cloud to Waymo. That is ground zero for us, making sure both for research and developing the models that we have enough compute needed. Beyond that, we have these independent businesses which each have their own long-range plans. They’re using those long-range plans to project forward, and we plan for compute for all of them. The cloud team, they have a business and they have a projection and they have demand signals. They have committed contracts. With that help, we can plan on a long-term compute trajectory for them. We’re fortunate to have this wide set of opportunities. There’s no company in the world which isn’t making trade-offs at any given moment in time. But the first principles we operate on is making sure we have enough compute to invest in the frontier, and that’s non-negotiable, and everything else works on top. Remember, we are building our own compute, our own data centers. There are long lead times. The compute we have today is a function of decisions I made in 2024. Even if you think back to a year from today, last year the agentic coding takeoff hadn’t happened. Today, obviously, when you sit and look at anything, everyone is short—not just us, other companies—and everyone wishes you had two times what you had planned for. That comes with the territory. At any given moment in time, we are planning for what we would need two to three years out in the future. You’re going to find some decisions you make may look suboptimal two years out, depending on where the state of the art is. I don’t spend too much time thinking about it. The way I think about compute allocation is making sure Google DeepMind, both for research and frontier model development, is cutting edge in terms of compute access. After that, we plan for all our businesses. It’s great we can provide our compute to other people too, like we have done. It allows us to scale that up, get real tests for it in the widest application set possible, test it in the real world, and make sure we are at the cutting edge of technology at every layer of our AI stack as well. Ben Horowitz relayed on a podcast that Andy Jassy had said compute was sold out until 2028. Are you seeing the same thing? If today a customer is coming out and asking us for compute in ’27, that is incredibly difficult to do. Even last year we went out and leased some short-term compute to accommodate some customers and help them out. But yes, we are all dealing with similar timelines. Every major technology shift seems to involve overbuilding, from railroads to fiber. Do you think that’s happening here? It depends on the time frames you’re looking at. This level of build, if I take a long enough time frame, it makes complete sense to me. We are at early stages of the most consequential technology ever, and this is not like fiber. It’s not dark. We are using it. We are capacity constrained on a daily basis. The applications, the underlying technology is improving rapidly. But running a company, the way I work through the moment is you look at your demand signals. You look at your committed revenue opportunities that you’re getting in the form of contracts. You balance that with the underlying technical capability, or internally you balance how we are using it internally, the model trajectories. Based on all that, you take a disciplined approach—a confident, aggressive, but disciplined approach—to plan out compute. But when you talk about the broader economy as a whole, you’re having people operate with various levels of risk. I focus on what I can control as a company and plan through it. If your question is, can there be an overbuild? Yes, because you can have some constrained components. Remember, what matters is a set of everything you need to build out that compute. Memory, electricity, electricians…. Land. Like 2,000 components or more. You could get constrained at some point. There could be things which cause bottlenecks, and then you could have overbuilds. There could be external shocks to the economy. There are macro risks. All that is possible at any given time. But am I confident that the compute we are planning for, we will have uses for it given enough time? Absolutely, yes. I was in the room when you said AI would be as profound as fire and electricity. If you believe that, shouldn’t you give researchers the biggest possible computer, even at the expense of today’s cloud profits? We’re all doing some form of that. Our cloud business is at any given moment in time forgoing short-term revenues because we are making compute trade-offs to support frontier AI development. What you’re seeing is true already. But you want to be thoughtful. Even our AI researchers have some constraints, which are also good. That’s what inspires creativity. That’s what helps you come up with a more efficient algorithmic breakthrough, maybe to use less memory. What we are committed to is to invest what’s needed to stay at the frontier. OpenAI President Greg Brockman told me a few months ago that the ideal AI product might have no interface at all—just as natural as conversation. What do you think the interface looks like in the age of AI? Computing historically has been humans adapting to computing. With AI, it’ll be the other way around. The technology will adapt seamlessly to how humans live and go about their day-to-day life. You want to be able to speak to that AI when you want to. It has context. It can be persistently running somewhere, give you continuity. You’re not dealing with a lot of complexity. It’s natural. All that makes sense to me. That doesn’t mean it won’t have any interface, however. Nowadays, even on my phone, Gemini voice models are so good. I can just press the microphone button. I can just ramble. That’s why we call it Rambler. I can ramble. It intelligently corrects with my personal context. Even my friends’ names, weird names, it gets it right, even if it first guesses it wrong. It’s amazing. I’m not going in, backspacing, editing once I finish saying it. It’s remarkable how good it’s gotten. But there are going to be moments you will need an interface. You may want to entertain yourself. You may want to watch a video. You may ask your AI to create something, and you want to watch it, consume it. The way I think about it is, you want it to be as seamless and as natural as possible, and that you are not doing the hard work adapting to whatever the interface is. The technology is adapting to how you’re working at any given moment in time. When I asked Sam Altman about competition from Google, he argued that OpenAI was building AI-native products and implied Google was bolting AI onto existing ones. Do you accept that framing, and can a “bolt-on” compete with an AI-native product? First of all, it’s a very dynamic moment. We are all pushing each other. OpenAI has pushed us as a company too. This is not just we are doing something to them. Far from it. Yes, Google went into its own version of a code red too. In the AI moment, all of us have to respond with varying levels of urgency at any given moment in time. We are constantly building new products as well. Gemini was built from the ground up as a product—the Gemini app I’m talking about. Or tomorrow, as all of us are adapting to agents, are you bolting agents onto ChatGPT, or are they building it natively? How would you think about that? You can always put these questions back. I don’t think of it that way. I think of it as you want to build world-class product experiences that are AI-native and they are winning. Some of it is adapting products, some of it is rethinking new experiences. They will all be part of how you approach this moment. How do you expect to make money from these AI interfaces? If an agent recommends a product and I buy it, should it take a cut of the sale, even if that’s disclosed? I’m going to put enterprise aside. Consumers are going to increasingly value this technology. Take software developers today. Developers are willing to pay for these coding models because it really helps make them more productive. Nobody would want to code the way they were coding three years ago, or even the same time last year. When you’re delivering value, there’s value to be had. I’ve never seen examples where consumer products really scale to billions of users, where people are using it intensely, and haven’t had monetization to go along with it. If you’re providing value to people, there is inherent value to be had. The nature of it may change. You’re asking a scenario in which agents are doing stuff for you. At the end of the day, you will have commercial needs. The agents need to fulfill those commercial needs. The providers on the other side are going to see value in that. There’s going to be some mechanism by which it gets fleshed out. That’s what an efficient system does. But what matters is that you’re building those world-class consumer experiences. I promised myself one product feature request: Can you make Gemini-powered search in Gmail as good as searching Gmail through ChatGPT? What’s going on here? One of the exciting things we announced recently is the unified Gemini agent for enterprise, and we’re bringing that to consumers. These are persistent agents which run 24/7 in the cloud. We’re spinning up a micro VM effectively for every query, and from the runtime, the harness, the model—it’s all a headless service. It’ll work everywhere, and it has your context with you. You can address this agent whether you’re in Workspace, in Search, in Gemini. Agents are going to drive this unification from a customer experience, which people are looking for. When you have a broad product portfolio and you’re moving forward fast, there are always areas where you’re moving forward faster than some other areas. That’s valid feedback. But we are putting a lot of time and effort into making sure, as we build out our agentic experience, that unification and that capability works well across all these services. People have asked where Google’s equivalent of Muse or Dots is. Do you anticipate creating a new product to compete with them? Our goal with Gemini is to be a personal, proactive, and powerful assistant. That’s what we are working toward. You saw us describe the early versions of it at Google I/O this year with Spark. Effectively, already, if you open the Gemini app as a Pro or an Ultra user, you see Chat and Task, and they start giving you that agentic experience. With Argon now, that capability is dramatically accelerated. What we announced recently in our enterprise event, we are bringing that to consumers as well. Soon in Gemini, you will be able to ask something, and it’s a continuum. You don’t need to think about whether you’re asking a chat or an agentic task. It’ll just automatically do the right thing, and it can work persistently in the background securely and get things done for you. Already, empirically, when I look at what’s coming into Gemini in our app, last year you would have tagged them and most of them would have been chats. But a steadily growing percentage is already agentic tasks in Gemini. Yes, we are going to bring personal agents to our consumers. It’ll work in the Gemini app. It’ll work in Google Search. It’ll work in Gmail. You can address it, call it, etc. We want to make sure we’re getting it right. It’ll be powered by Argon, and we are very excited to bring it to our users. Google has invested and committed tens of billions of dollars to Anthropic. What do you see in Dario Amodei that makes you want such a deep investment and partnership? Very early on, when we looked at the founding team of Anthropic, it was very clear to us this was a world-class team which was involved in some of the earliest GPT models. We were quickly able to assess that they would be at the frontier. We’ve always tried to identify and partner with companies which would be at the cutting edge of technology. We’ve done that with SpaceX. We’ve done that with Anthropic. That’s what we saw, and we had a chance to partner. Is your partnership with Apple to license Gemini for Siri expanding, contracting, or standing still? For both us and Apple, that partnership was about translating the capabilities of AI into better experiences for users. I thought Apple has done a phenomenal job of incorporating—they customize these models and build it into Siri—and it’s been great to see the reception. We are committed to doing more. You’re going to do more with them? There’s nothing specific to comment, but it’s a partnership. It’s working well. What have you learned so far about the feasibility of data centers in space? It’s a very exciting project. Projects like this remind me—I still recall that moment in the desert for the DARPA Grand Challenge, watching the Stanley car compete in it, which ended up being the basis for Waymo. With all these technologies, you’re at a moment in time where it looks a bit like sci-fi, but 10 to 20 years down the line, it’s actually working in a meaningful way. I feel the same in space. We’ve always believed in betting big on these technologies. Quantum has been a great example of it. We’re probably 10 years into the journey. Waymo, we are almost 20 years into the journey. Space, we are two to three years into the journey. It’s clear to me that it’s an approach which will make a lot of sense in the future. Being in sub-synchronous orbits with 24/7 access to the most abundant form of energy we have access to will make a lot of sense for this exponential increase in compute that we are going to need. Having said that, this is a challenging project. The team has disciplined milestones. It was a super exciting milestone to send the TPUs up to space. Sometime, as we are speaking, we are expecting a “hello world” moment from them. I look forward to seeing that happen, and we’ll continue investing here too. You previously said 25% of the code written within Google was written by AI. What’s the number today? That number in our world today is meaningless in the sense that every part of what we now touch is influenced by AI deeply. What we are tracking with Argon, for example, it’s been a point of inflection. People are using Argon to really go tackle complex projects, including projects which affect millions of lines of code. In the four weeks since we have launched Argon, we have seen a 30% increase in CLs, and more importantly, almost a similar increase—a one-third decrease in rollbacks. People are writing code more productively and better code with it. We are tracking it with newer metrics now. How are you dealing with opposition to data centers? If you can’t build in the U.S. the way you want to, where else do you go? We’re very committed to building our data centers very responsibly. When I look at our largest data centers in some of the communities where we have been the longest, I visit them, I talk to people. There are people who come and tell us their kids have gone to college because of our investments in these communities, and we’ve grown in those communities. To me, this is about doing it right. Engaging with the communities early, doing it transparently. We are committed to making sure we are bringing incremental power, and we cover 100% of the cost of our power needs for these projects. It’s understanding the impact on neighborhoods and making sure you’re working thoroughly to address them. Done rightly, you can continue making progress. But we are constantly taking feedback into account. I do worry not everyone operates the same way. People have been surprised by data centers showing up in their backyard. Those are all important. As an industry, we should continue to do better. But part of the reason, in a 20-year time frame, we are thinking about things like data centers in space is making sure we are able to put these data centers where maybe the trade-offs will be different. What unproven area of AI is Google working hardest to crack? Is it continual learning—getting models to keep updating after release—or something else? There are multiple parallel paths. You saw with Argon we have a million output tokens. With our Gemini models early on, we had a one-million context window on the input side. These are all examples of breakthroughs we are constantly working on. How do you think about synthetic data? How do you think about continual learning? How do you think about more efficient architectures? We do dedicate a meaningful percentage of our compute to these research bets too. It always goes hand in hand, including our world models. Those are all bets we are placing. Sometimes, as part of these bets, there are ideas that emerge which just make their way into the mainline models. Maybe we don’t fully talk about it, but we have multiple steps in these areas. Working in technology requires optimism that building something will improve the world, but those bets can leave real consequences when they fail. I recently returned to a town in El Salvador where enthusiasm for Bitcoin had given way to unfinished construction and something resembling a ghost town. How do you maintain the optimism you spoke about in your Stanford commencement address while accounting for those risks? I genuinely have a deep conviction that improving technology and improving access to technology makes people’s lives better. It’s been my personal experience in my life, out of which comes that deep conviction, optimism. But with that, I’ve always felt humanity’s journey is always about working hard to harness that technology to make it work well for society. The optimism is not an optimism that you just build whatever you want, whenever. You work hard to make sure you’re building technology in the service of humanity, to make the human condition better. That’s got to be the defining goal as we are approaching it. With that framework, I’m very, very optimistic. We are in very early stages of the kind of progress that we are going to see on many areas. We’re all talking about areas like healthcare. A lot of the progress is ahead. The ability to make things in a fundamentally cheaper way on many things where things have been getting expensive for a long, long time, discovering new materials, etc., which really bend the curve—it’s tough to step today and understand the impact of all that, but they’re going to be very profound. I’m very optimistic, but we have to be rightfully concerned about all the ways this technology can go wrong. Sundar, I know you don’t do many of these. Thank you. Thanks, Alex. Appreciate it.
4 Cheaper Alternatives To Expensive Smart Home Sensors
Smart home sensors can be expensive, but you don't have to break the bank to start automating. Cheaper alternatives include sensors from IKEA and Sonoff.
AI video tools like Slopcannon create new competition for Hollywood
AI video tools are getting better fast, giving Hollywood cheaper production -- and a flood of new competition.
Can California help lower the cost of GLP-1s? Here’s what lawmakers are proposing
An LA-area Democrat wants California to partner with drug manufacturers to make weight loss drugs cheaper — drawing on her own experience paying out of pocket.
Sony could release a cheaper version of its WH-1000XM4 headphones, according to leaks
Sony could revive an older headphone model at a cheaper price.
Politics & Power: Big opportunities, real risks, ethics & oversight as AI reshapes the legal industry
AI can make legal work faster, cheaper, and more accessible, but it doesn't come without risk. The organizations that come out ahead won't just be the ones that adopt the best tools. They'll be the ones that pair those tools with clear oversight, honest communication with clients, and a genuine commitment to getting it right.
Beckett Reopens Base, Standard Grading While Website Remains Offline
Beckett’s cheaper grading tiers are accepting cards again, but collectors still won’t find the normal submission process waiting for them online. The post Beckett Reopens Base, Standard Grading While Website Remains Offline appeared first on HEAVY.
State Fair of Texas brings lower prices, new attractions, more value options
From cheaper rides and games to over 200 food and drink options priced at $10 or less, the fair is putting a bigger focus on affordability.
SCOTUS grants Republican appeal on TV ad rates before midterms
The order clears a legal roadblock to political parties getting cheaper TV ad time ahead of the midterm elections
She Pays $260 a Month for Plan G. The Same Insurer Sells a Cheaper Version of It, and Getting In Means Clearing Its Health Questions.
She called her insurer's sales line out of curiosity and heard a price for her exact plan that was materially lower than what showed up on her renewal. Getting from one number to the other means clearing health questions she never knew would matter this much.
With new Macs, Apple aims to take on Microsoft, Nvidia in a rush to lower AI costs
When Apple's new desktop computers start shipping Tuesday, the company's executives will make an unusual pitch to corporate buyers: They are cheaper than renting data centers.
OLED Screens Are Finally Defeating the One Thing Holding Them Back
The prettiest displays available may be getting cheaper.
Amtrak business class vs. coach: I liked the cheaper train ride
I rode business class and coach on an Amtrak train between New York City and Rochester. The perks in business class weren't worth the extra money.
5 Cheaper Android Phones To Try Instead Of The Google Pixel 11
Previous-generation flagship phones are still excellent and now they go for cheaper. That includes the Google Pixel 10 and Samsung Galaxy S25.
What’s New at the State Fair of Texas
Cheaper tickets and Midway rides, an indoor main stage, new kids’ areas, fresh exhibits, Big Tex’s new look, and the return of the Old Mill Inn are among the changes waiting at Fair Park.
Apple’s ‘new’ polishing cloth is the same except $10 cheaper
The same polishing cloth, but at a lower price.
You might not miss what Nikon removed from its cheaper full-frame Z5 IIC
The 24.5-megapixel camera ditches some of the Z5 II’s premium features but shaves $450 off the original’s price.
Supreme Court rulings change the economics of political party spending
A Supreme Court ruling last week allowing political parties to continue accessing cheaper TV and radio advertising may give a boost to Republicans as campaigns
This cartridge-playing Game Boy clone is smaller and cheaper than Analogue’s Pocket
The FPGB Mini is only slightly wider than the original Nintendo cartridges it plays.
Scientists reveal why some workout recovery tricks can backfire
Ice baths, massages, compression garments, and recovery supplements promise faster muscle recovery, but research suggests their benefits are often overstated, and some techniques may even undermine long-term gains. Scientists say the most effective recovery methods are far simpler and cheaper: getting enough sleep, eating well, and giving muscles time to rebuild.