SCROLL NEWS / DISCOVERY
Search headlines
Find the stories shaping the conversation.
Results for " simulation " 194 found 🔀 AI-powered Shuffle
From Materials Simulation to Experimental Astronomy, New NVIDIA AI Software Unlocks Scientific Discoveries
NVIDIA CUDA-X libraries, microservices and reference code accelerate AI for science.
STCC breaks ground on $55M health school move
Springfield Technical Community College held a celebration this afternoon for a groundbreaking ceremony to mark the relocation of its School of Health and Patient Simulation.
Fairway Bays virtual golf simulation business to open on Father’s Day
Business located in Tioga Town Center.
There aren’t enough cadavers to teach medical students. Surgical Science wants to help
The company makes surgical simulations that teach students how to operate on humans.
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.
Sutter Health launches California’s first full-scale mobile simulation lab
Officials said the initiative is designed to strengthen teamwork, communication, critical thinking and evidence-based practice among interdisciplinary care teams.
Science teachers learn about STEAM at Jump Simulation Trading & Education Center in Peoria
PEORIA, Ill.-Science educators from across Illinois got a firsthand look at innovative STEAM learning opportun...
Science educators explore STEAM lessons during immersive visit to OSF Jump Simulation & Education Center
Key Takeaways:Science educators from across Illinois explored hands-on STEAM activities at OSF Jump Trading Simulation & Education Center, gaining classroom-ready ideas and resources.OSF HealthCare highlighted free educational tools and activities that inspire student interest in health care and STEAM careers.Educators discovered new engagement ...
Patriots 2026 season simulation: A.J. Brown's stats, final record and more
How will A.J. Brown fare in Year 1 with the Patriots? Will the star WR help New England to another stellar season? Here's Strat-O-Matic's 2026 simulation.
Tiny X-ray telescope could unlock the Moon's hidden chemistry
A lightweight new X-ray telescope could finally give scientists something they’ve never had before: a complete chemical map of the Moon. Researchers used detailed mission simulations to show that a compact telescope orbiting the Moon could identify key elements across the entire lunar surface, helping reveal how the Moon formed and evolved.
Scientists identify the origin of noise in spin qubit quantum processors
Charge noise arising from two-level fluctuators is considered to play a key role in causing qubit frequency shifts in silicon spin qubits, resulting in deteriorated gate fidelity. Higher temperatures can improve gate fidelity, but the microscopic origins of this effect and of qubit frequency shift have not yet been established. Now, using statistical simulations, researchers have clarified the parameter regimes under which gate fidelity can be improved and the potential origin of qubit frequency shifts.
The cult of objectivity has broken consciousness science
With 22 competing theories and no consensus in sight, the science of consciousness is lost. In this exclusive IAI interview, Berkeley psychedelics researcher and bestselling author of a new book on consciousness Michael Pollan insists that even though consciousness lies beyond traditional forms of scientific investigation, we should resist becoming dualists who see it as something radically different from ordinary stuff in the world. Instead, he makes the case for reimagining science itself: it must stop reaching for objective descriptions of physical reality, and instead embrace subjectivity as the indispensable core of all inquiry. Alasdair Craig: How did your experiences with psychedelics get you interested in consciousness?Michael Pollan: One of the things that happens on psychedelics is that they foreground consciousness. What formerly seemed perfectly transparent and ever present—and yet, non-existent, as water is to fish—suddenly you see it. The metaphor I use in the book is that psychedelics sponge the windscreen. Suddenly you realize: wow, there’s a windscreen, there is this phenomenon that’s mediating my relationship to reality. This is not an unusual realization on psychedelics. I think a lot of people are called to think about consciousness after psychedelic experiences. So that was the proximate inspiration for A World Appears: A Journey Into Consciousness (2026).And then there was a more specific way in which psychedelics influence the book. That was a particular experience I had on psilocybin. I had the sensation that the plants in my garden were conscious and returning my gaze, with their own point of view and agency, and I got curious as to how to evaluate an insight like that. I mean, what’s the truth quotient of insights you have on psychedelics? And I was reading William James—who, of course, wrote beautifully about mystical experience—and he said, we can’t judge the metaphysics here, but we should pay attention to two criteria: (1) How useful is it to think this is true? He was a pragmatist, of course. And (2) can we subject it to other ways of knowing to validate it? Because maybe we should treat these insights as hypotheses. And that’s kind of how I approach psychedelics, as a great generator of hypotheses, but they need to be tested against other ways of knowing. And so that led me down this long path exploring the science of plant intelligence and plant sentience.AC: I love your phrase, “smudging the windowpane of conscious experience.” So do you now start to think, actually, even when I’m not on psychedelics, I’m just seeing a screen all the time, which is not reality, but rather is my conscious experience, behind which reality is hidden?MP: Well, you know, not when I’m brushing my teeth and making breakfast! But yes, in meditation, certainly. I mean, one of the things you realize is that my experience happens to be this way now, but it could be another way. And many things can affect it. My morning coffee changes the screen in various ways, sharpens certain things.___The only tool we have to understand consciousness is consciousness.___There’s this wonderful metaphor in Aldous Huxley’s book, The Doors of Perception, where he thinks we channel consciousness, and that what the brain does is edit it, and there’s a reducing valve. And psychedelics open the reducing valve wider. I don’t know that I’m ready to buy into the idea that consciousness exists outside brains, but I do love that metaphor because there are times when you are more conscious than others. I know I’ve switched from window pane to valve, but you know, a camera aperture can open wider or narrower. And I do have a sense of that. I have a sense of when I’m being less conscious, and when I’m being more conscious, by which I mean, you know, more present, taking in more sensory information. So when you start thinking about consciousness as a matter of degree, that changes things also.I think when we scroll on social media, obviously, we’re conscious. But minimally so, because we don’t have to generate our own thoughts. We’re basically taking these little hits from other people and thinking their thoughts.The process of writing this book made me kind of relativize everyday consciousness as just one type, which is exactly what William James said. Normal consciousness is one type, separated by the flimsiest of screens from the other types.AC: If we stick with the metaphor of the smudge on the window that you get when you take psychedelics, it seems like we’re saying that consciousness has become like an object that you’re looking at, right? You’re looking at the smudge and that just is consciousness. But sometimes we’ll talk about consciousness as a state that you’re in, rather than an object you can perceive. You’re either conscious or you’re not conscious. So it’s difficult to really know what sort of thing we’re talking about when we talk about consciousness, right?MP: Yeah. When you say that consciousness is an object, what’s the subject? Consciousness also? The only tool we have to understand consciousness is consciousness. So we’re in a closed loop and when we get a perspective on consciousness, we’re not achieving any distance from it. That’s one of the problems with studying consciousness. Science is predicated on an objective, God’s eye view of things, and you never get that. There’s no such thing. SUGGESTED VIEWING The brain filters consciousness With Àlex Gómez-Marín But, yeah, there is this sense, I think, that consciousness is something you can approach. Certainly, when you’re writing and trying to recreate a conscious experience, it’s a fiction, but you’re standing back from it, and looking at it. But there’s no real subject-object distinction.AC: That’s so interesting, because when philosophers start talking about consciousness, they often talk about what they call “qualia”, which is their word for the felt experience of, say, the color red. And I think that way of speaking often makes us think of the qualia as an object, which you can see out there: the redness of red. But, as you say, we also want to collapse the subject-object distinction when we talk about consciousness. So there’s something strange going on here…MP: Oh, yeah, there are many strange things going on! *AC: Let’s talk about the varieties of consciousness. Apart from perceptions, there are, pains, emotions, thoughts, and so on. What is it that holds all this diversity together so that they all count as instances of consciousness? What is it that they all have in common?MP: Yeah, that’s a really good question. I mean, there is sensory information, right? Which we perceive coming from the world outside ourselves. And then there’s internal information, which is what’s called interoception. And this is very important too. This is where feelings come from—what are sometimes called homeostatic feelings—that are intended to get the brain to keep us in these narrow ranges of temperature and glucose and so on. That’s a very important part of consciousness. And according to some researchers, who I think are probably right, this is where consciousness begins, with these very primal feelings, like hunger and thirst and warmth and itch, and things like that. And then there’s this third thing, which is our thinking about thinking.And what holds all this together is, I think, the sense of self. Consciousness generates a self: it is probably an organizing principle. All this stuff is happening to the self. It’s a construct, I think, but it is very useful. It can connect all these different streams of consciousness to a single being or a place, metaphorically speaking.* AC: The first half of your book is about the science of consciousness. Is consciousness just another thing, like black holes or the origin of life, which we don’t yet understand, but which we should have every hope that science will one day get to the bottom of, or is there something special about consciousness that makes this unlikely?MP: It depends how you define science. As we define science now, it’s going be very difficult. Science since Galileo has had a set of rules or assumptions, which is that the proper remit of science is to deal with the objective, quantifiable, third-person world. And the physical sciences have done amazing work under that set of limitations. Galileo—for tactical reasons, not for philosophical reasons—said science should not deal with the soul, which at the time was what consciousness was considered to be. So what is subjective, what is qualitative, what is first person, we’re going to leave to the church, he said. And this was a good agreement, and it probably kept many scientists from getting burned at the stake. But it leaves science in an awkward spot, now that it wants to venture into that territory of qualities, of first-person experience. And I don’t think it has the proper tools. I think it will have to add to its toolkit in some way, and find a way to bring in phenomenology, or lived experience and introspection, which science is usually dismissive of.The challenge—we alluded to it earlier—is that the only tool you have to study consciousness with is consciousness. And what I mean by that is, the whole scientific enterprise is a manifestation of consciousness. As much as scientists like to flatter themselves with this idea that they attain a godlike perspective on what they study, when it comes to consciousness, there’s no such thing. And it’s fine for all the things you can step outside of and study from that perspective. But two areas it won’t work are consciousness and cosmology.___It could be that we live in this anomaly of not believing nature is conscious. And maybe psychedelics are teaching us something we’ve forgotten.___With cosmology, too, you have to study the universe from inside the universe. It’s not that cosmologists haven’t made progress. We know a lot about the age of the universe and expansion and things like that. But we can’t penetrate what happened before the Big Bang. We just don’t have the tools because we can’t stand outside.The fact that there are 22 full-scale theories of consciousness tells you all you need to know: we’re kind of lost. And I was surprised to find that we are more lost than I thought we were when I started this. I thought that there’d be leading theories, or that we had experiments that we could conduct that would nail some of this down. And the fact is, we don’t.So, I don’t know if science can solve it. But, of course, we can change the rules of science. A really influential book for me was a book called The Blind Spot, by two physicists, Marcello Gleiser and Adam Frank, and a philosopher, Evan Thompson. It’s a reminder that science is a social practice. It’s organized in a certain way and it can be organized in a different way. But it has this huge blind spot at the center of it, which is its failure to recognize that it is a form of consciousness itself, and it doesn’t have a good way of taking account of lived experience.AC: On the one hand, I think you want to resist traditional dualism, which would say that consciousness is something magical, something supernatural, right? But on the other hand, you’re accepting that there is this division in reality between the stuff that you can investigate objectively and measure, and so on—the stuff that science has been good at finding out about—and consciousness, which, for some reason, you can’t objectively measure. Some would take this to suggest that the dualism is actually right. Is there a way of reconciling these two strands of your thinking?MP: To call it magical is full of value judgement. I mean, you could just say immaterial. The interesting thing is, I think, that intuitively we’re all dualists. And I don’t know whether that’s because we’re children of Descartes and other cultures are less dualist than we are. Or it may be because that’s how consciousness feels. It feels different than the body. It feels separate from the body. But I don’t think we should trust that intuition.AC: So let’s think about ways that science might be changed to get at consciousness. I suppose the starting point would be the way that you begin the second half of your book. You have this nice line from William James, where he says, “We now begin our study of the mind from within.” Could you give us some sense of the ways of studying the mind from within? What does this involve?MP: I don’t know what that science looks like. But one of the points I’m making is that it depends on the humanities. I found that I learned as much, if not more, about consciousness, from reading novels and poetry. There is a sense of the fullness and complexity of consciousness that you don’t get in any scientific account. I interviewed this stream-of-consciousness writer and I approached her as an expert on consciousness, and she is. So I think that dialogue is really important.I’m reading Ben Lerner’s new novel, Transcription. Just read the first page of this novel: it’s a man falling asleep on a train, going to conduct an interview, and there is the experience of being in the train, facing backwards, there’s the view outside, but there’s a reflection of himself through which he’s seeing outside. There’s the memory of dropping his kids off at school. All this is going on in one moment of consciousness. Scientists aren’t going anywhere near that. But I think it’s useful for them to think about. That’s where we’re trying to go: like, how does such a layered experience happen? You know, it also makes you wonder, could a computer ever do this? And it gives you some doubt.*AC: Before we get to computers, perhaps we could talk briefly about plants. How did you come to feel that they are sentient?MP: Well, it was a psychedelic experience, and it’s not unusual on psychedelics. In fact, there’ve been studies about how people become animists during psychedelic experience. Their belief in animism arises after a single psychedelic experience.You know, most scientists would say this is all a myth. But we have only 400 years of the age of enlightenment and science. Before that, we were animists, and if you go around the world, most indigenous cultures are still animists, and children are animists, by and large, until they go to school. So it could be that we live in this anomaly of not believing nature is conscious. And maybe psychedelics are teaching us something we’ve forgotten.As I said, I wanted to test this idea, and see whether there was conventional scientific evidence for it. And there’s fascinating work going on in a field that calls itself—in a rather cheeky way—plant neurobiology. These are people who fully realize there are no neurons in plants, but they’re trolling conventional botanists! And they’re doing experiments that are demonstrating that plants can learn and remember, which is kind of amazing: how do you store a memory without neurons? They’re finding plants that can solve problems, that can hear and see. In fact, they have 20 senses to our five or six. For example, a bean plant looking for something to twine, and given a pole within eight inches or so of it, knows where that pole is and will aim its spiraling reach at that pole. What is that about? Is it bouncing signals off of it?One of the spookiest things is the fact that the same anesthetics that put us out put plants out. You might think that plants are already out, but they’re not. They’re just slow. If you take a plant with a fast behavior, like a Venus flytrap, and give it an anesthetic, it will stop reacting. So you have to ask yourself, what is lost when the plant is anaesthetized? Some state of alertness, of awareness, of its environment, of reactivity. I’m not prepared to call that consciousness, but maybe it’s sentience—which is a simpler form of consciousness, and involves the ability to sense what’s going on in your environment and assess what’s good and bad for you and act appropriately. That may be a universal quality of life.AC: Let’s turn from plants to AI. You point out that LLMs are trained on the internet rather than direct interactions with reality. You put this really nicely when you say the internet is not the world so much as a shadow cast by the world. So LLMs are living in this kind of shadow world, or in Plato’s cave, as you put it. What’s the significance of this for AI consciousness and moral consideration? And, to recall the beginning of our conversation, if we’re stuck behind a veil or screen of consciousness, are we also stuck in a kind of shadow world?MP: I don’t think they’re the same shadow worlds because our world includes nature and other people. And even if it’s mediated in some ways... There is a reality out there, and we are in relation to it, through consciousness. There’s friction with it. What’s missing from the LLMs is that friction with the real world. There’s more to the world than human representation, which is what they’re dealing with. And that means that so far, AI is a cultural technology for combining thoughts we’ve already had. SUGGESTED VIEWING Cracking the code for thought With Roger Penrose, Sabrina Gonzalez Pasterski, Max Tegmark, Güneş Taylor Genuine creativity is more than collage. A lot of what AI does now is tantamount to collage, putting this with that and you know, that’s part of art, certainly. But there are also genuinely new ideas that come into the world that don’t come in through manipulating the representations we already have. So it remains to be seen whether AIs can be truly creative in that more radical sense.But my arguments against AI becoming conscious have to do with the fact that the idea of AI consciousness is based on this faulty metaphor that the brain is a computer. There are dimensions in which the brain is like a computer—it certainly does its share of computation. But computers have this sharp distinction between hardware and software, and brains don’t. In a brain, hardware and software are identical. Every memory is a physical change to the brain. Brains are not interchangeable in the way computers are. So this idea that you could take the consciousness out and somehow run it on another substrate, a silicon substrate, seems foolhardy.Anyway, it’s a more extended argument, and I may be wrong. And when we move to computers that aren’t based on silicon—neuromorphic computers and things like that—all bets may be off.AC: Regarding your point about LLMs being stuck in the world of our representations—the internet or whatever—and not being able to interact with reality, are these representations, the world of the internet, themselves part of reality? We’ve created the internet, but it’s still a real thing that exists. And so while the LLMs may be trapped within that part of reality, they’re nevertheless within reality and interacting with it and so on. So it’s not obvious that there’s a barrier there to them becoming conscious and being creative.MP: Yeah, I mean, if our consciousness is built out of human culture, which is what we’re talking about with all those representations. But I tend to think consciousness comes a bit more out of nature and friction with the natural world, but...AC: So there’s a fundamental divide between the natural world and human artefacts?MP: And we lose track of that, and especially people in this computer world. You know, the people who are creating these AI systems, my guess is they’ve been into video games since the age of three, and they didn’t go outside a lot. And their own contact with nature is somewhat limited. I could be wrong, it’d be a very interesting thing to ask, all the wizards of AI, whether they like to go fishing and hiking and do things like that. Or were they on computers from a very early age? I interviewed Blake Lemoine, the Google engineer who came to believe that LaMDA, the Google LLM out at the time, was conscious and deserved personhood. It was very clear that he didn’t see a distinction between this bubble in which he lived and the real world. And I think we’ve all been distanced from the natural world, and this is taking us even further away. But it’s still there, you know, and we still have bodies, and we’re still mortal, and we don’t live in a simulation.AC: I think that’s a great place to finish. Michael Pollan, thank you very much!
FDA sees potential in new AI software for monitoring heart failure patients
The software-only platform helps clinicians keep track of patients when they are at their most vulnerable.
ThedaCare's new simulation lab offers cutting-edge training space for its healthcare team
ThedaCare is preparing to welcome its first cohort of residents to train in a new, state-of-the-art educational facility.
Hyperreal Politics: When the Simulation Becomes Too Absurd to Satirize
My dialogue with Claude continues as we reflect on civilizational decline, democratic erosion, and the capture of shared
Massive supercomputer simulations unlock cosmic magnetic mystery
Scientists used some of the most advanced plasma simulations ever created to uncover how the universe builds enormous magnetic fields out of turbulence. The discovery could reshape our understanding of stars, black holes, neutron star collisions, and dangerous solar eruptions.
Hyperreal Politics: When the Simulation Becomes Too Absurd to Satirize
The foreign policy playing out today in the West is so preposterously self-destructive it would be rejected as implausible fiction
Science behind sobriety field tests
Albany High students in Elizabeth Lopatniuk's health classes participated in a dynamic lesson about the effects of alcohol on the body. After taking an in-depth look at the science behind why and how alcohol slows the activities of the brain and central nervous system, students learned about how a variety of factors can impact the level of intoxication and how blood alcohol concentration can impair one's ability to operate a motor vehicle. Albany High Student Resource Officer Nick Schuler led students through an overview of field sobriety tests that are routinely performed during a traffic stop when an officer suspects a driver may be under the influence of alcohol. Officer Schuler explained the three elements of a physical sobriety test, including walking a straight line with turns and checks for motor skill patterns, standing on one foot to check for balance and following a moving stimulus with eyes only to check for Horizontal Gaze Nystagmus (HGN), an involuntary, rhythmic jerking of the eyes as they look to the side. Students volunteered to try walking a straight line both with and without alcohol impairment simulation goggles. Officer Schuler pointed out particular behaviors and patterns that would lead to a failed field sobriety test during the simulation. As prom and graduation festivities approach, we applaud our Albany High staff for empowering our Falcons with valuable lessons so they can make wise and safe choices.
Science behind sobriety field tests
Albany High students in Elizabeth Lopatniuk's health classes participated in a dynamic lesson about the effects of alcohol on the body. After taking an in-depth look at the science behind why and how alcohol slows the activities of the brain and central nervous system, students learned about how a variety of factors can impact the level of intoxication and how blood alcohol concentration can impair one's ability to operate a motor vehicle. Albany High Student Resource Officer Nick Schuler led students through an overview of field sobriety tests that are routinely performed during a traffic stop when an officer suspects a driver may be under the influence of alcohol. Officer Schuler explained the three elements of a physical sobriety test, including walking a straight line with turns and checks for motor skill patterns, standing on one foot to check for balance and following a moving stimulus with eyes only to check for Horizontal Gaze Nystagmus (HGN), an involuntary, rhythmic jerking of the eyes as they look to the side. Students volunteered to try walking a straight line both with and without alcohol impairment simulation goggles. Officer Schuler pointed out particular behaviors and patterns that would lead to a failed field sobriety test during the simulation. As prom and graduation festivities approach, we applaud our Albany High staff for empowering our Falcons with valuable lessons so they can make wise and safe choices.
New MatterChat Model Helps AI to ‘See’ the Language of Science
From writing emails to generating computer code, much of the artificial intelligence prevalent in our daily lives has succeeded by mastering one domain: text. However, this leaves a major blind spot in the physical sciences, where models depend on the high-resolution, three-dimensional data of the physical world, like the intricate lattice of atoms in a crystal. Delivering on the promise of using AI for science requires teaching these data-driven text models to seamlessly “talk to” physics-based models. Now, a new AI framework from Lawrence Berkeley National Laboratory (Berkeley Lab), called MatterChat, solves this problem by creating a specialized “bridge.” It connects the conversational power of a Large Language Model (LLM) with a physics-based AI that models “interatomic potentials”: the complex physical forces between atoms. The resulting system already significantly outperforms general-purpose AI tools like GPT-4 at predicting material properties, and the team hopes it can accelerate scientific discovery by serving as a robust research partner that provides grounded insights and generates step-by-step instructions for synthesizing novel materials. A paper describing this work was recently published in Nature Machine Intelligence. “Traditional simulations can provide the physical rigor required for materials science, yet their computational cost remains prohibitive for high-throughput screening. Conversely, while LLMs excel at rapid knowledge synthesis, they inherently lack the ‘structural vision’ to interpret materials directly from their underlying atomic coordinates,” said Yingheng Tang, a postdoctoral researcher in Berkeley Lab’s Applied Math and Computational Research Division (AMCR) and lead author on the paper. “MatterChat was built to solve this dilemma, empowering LLMs with a structural ‘vision’ that allows researchers to leverage their full potential for solving complex, real-world materials challenges.” Empowering language models to solve complex challenges in materials science To build MatterChat, the Berkeley Lab team drew inspiration from technologies like Vision Question Answering (VQA) and Text-to-Image (T2I) generation. In these tasks, AI must translate high-level text concepts into visual images, or vice versa. Doing so requires developers to build tools that “bridge” two fundamentally different forms of data. The researchers adapted this concept to the physical sciences. With MatterChat, they created a “bridge model” that successfully connects an LLM’s general knowledge with the deep understanding of the atomic-scale world encoded in scientific interatomic potentials. Until now, researchers using LLMs to solve materials problems usually had to feed them raw data files as if they were just strings of text. It’s like asking an AI to understand a complex 3D engine based only on a parts list: the LLM can read the names, but it can’t “see” how the atoms fit together in space. MatterChat solves this by training a specialized AI bridge model, pre-trained on millions of crystal structures and an LLM, to align the LLM’s representation of the world with the interatomic potential’s representation of the world. The bridge model can translate physical insights into a format the LLM can actually understand. By giving the LLM these “scientific eyes”— a scientific “inductive bias” in the terminology of AI — the Berkeley Lab team has transformed it into a robust research tool capable of providing grounded scientific insights into complex materials challenges, such as predicting thermal stability or analyzing electronic band gaps. “We think of atoms as living in a physical space, but from a machine learning perspective, they are just vectors living in some very non-trivially structured manifold in a high-dimensional Euclidean space; and, of course, the same is true for the sentences and paragraphs that we use to express our ideas about those atoms,” said co-author Michael Mahoney, Berkeley Lab’s Scientific Data Division (SDD) AI Initiative Research Lead. “The bridge model basically gets those two structures to ‘talk with’ each other.” As a proof-of-concept of this general approach, the team trained their bridge model on a dataset curated by pairing nearly 143,000 stable atomic structures from the Materials Project with their corresponding physical properties. This training data was automatically assembled using the Materials Project’s API and deliberately enriched with properties fundamental to microelectronics design — like formation energy and bandgap — allowing MatterChat to learn the complex patterns connecting a material’s atomic blueprint to its functional performance. To validate their model, the researchers benchmarked MatterChat against a suite of other AI systems, from general-purpose LLMs to other specialized scientific AI methods. The results show that MatterChat consistently outperformed its competitors across a range of tasks. The model was more accurate in classifying material types and demonstrated superior precision in predicting numerical properties. For example, it excelled at predicting a material’s bandgap, a property critical for designing new electronics from high-capacity energy storage to next-generation computer chips. “Our design is significantly more efficient because we don’t have to build a massive AI model from the ground up,” said co-author Zhi (Jackie) Yao, a research scientist in Berkeley Lab’s AMCR. “Instead, we take two powerful, pre-trained models — a structural encoder for materials physics and an open-source LLM — and use them off-the-shelf. The only component we actually train is the lightweight ‘bridge model’ that translates between them. It’s the difference between building an entire car factory and simply designing a smart adapter that connects a world-class engine to a world-class navigation system. This approach is not only computationally efficient, but also makes the system modular, so we can easily upgrade components or adapt the bridge for other scientific domains in the future.” Crucially, this modular design highlights exactly how institutions like Berkeley Lab and the Department of Energy are carving out a highly valuable niche in the booming AI landscape. Rather than competing with Silicon Valley tech giants to build ever-larger language models from scratch, the lab is focusing on the specialized connective tissue that makes commercial AI useful for hardcore science. Because the bridge model approach underlying MatterChat is forward-compatible, it is perfectly positioned to leverage these parallel tracks of innovation. As Mahoney pointed out, “We expect that industry will continue to develop improved LLMs, and we expect domain scientists and facilities will continue to generate new data. An important part of scientific machine learning is not simply to solve problems on today’s data, but instead to develop general methods that will be forward-compatible with orders of magnitude more data, whether from scientific domains or from LLMs.” According to Yao, the MatterChat project, which was initially developed and enhanced with funding from a Berkeley Lab Laboratory Directed Research & Development (LDRD) Program, will now expand its capabilities. In a collaboration with Fermilab, MatterChat is already contributing to a U.S. Department of Energy Genesis Mission project — called Accelerating eXtreme Environment Specs-to-Silicon (AXESS) — that aims to speed up the development of next-generation, high-speed, radiation-hardened detectors for challenging particle physics experiments by using advanced 3D integrated circuits (chiplets) and AI-driven data analysis. In addition to the LDRD support, the team also credits supercomputing resources at the National Energy Research Scientific Computing Center (NERSC), located at Berkeley Lab, with MatterChat’s success. “We are incredibly grateful to NERSC; this research simply would not have happened without access to the Perlmutter supercomputer through their AI for Science program,” said Tang. Wenbin Xu, a NERSC postdoctoral fellow at the time, was also a major co-author of the work, as was Benjamin Erichson, a research scientist in Berkeley Lab’s SDD, highlighting the benefits of AMCR-SDD collaboration on AI for science. ### Lawrence Berkeley National Laboratory (Berkeley Lab) is committed to groundbreaking research focused on discovery science and solutions for abundant and reliable energy supplies. The lab’s expertise spans materials, chemistry, physics, biology, earth and environmental science, mathematics, and computing. Researchers from around the world rely on the lab’s world-class scientific facilities for their own pioneering research. Founded in 1931 on the belief that the biggest problems are best addressed by teams, Berkeley Lab and its scientists have been recognized with 17 Nobel Prizes. Berkeley Lab is a multiprogram national laboratory managed by the University of California for the U.S. Department of Energy’s Office of Science. DOE’s Office of Science is the single largest supporter of basic research in the physical sciences in the United States, and is working to address some of the most pressing challenges of our time. For more information, please visit energy.gov/science.
Scientists finally solve the 100-year mystery behind tough tires
For nearly 100 years, reinforced rubber has powered everything from car tires to airplanes, yet scientists never fully understood why adding tiny particles of carbon black made rubber so incredibly strong. Now, researchers at the University of South Florida have finally cracked the mystery using massive computer simulations that took the equivalent of 15 years of computing time. They discovered that carbon black forces rubber to “fight against itself” when stretched, dramatically boosting its strength and durability.
iRacing for Apple Vision Pro Now Available
iRacing is thrilled to announce the iRacing Connect app for Apple Vision Pro is now available on the App Store. The app facilitates connections between the world’s premier motorsport simulation and Apple Vision Pro, bringing the thrill of sim racing to an exciting new platform. iRacing for Apple Vision Pro utilizes foveated streaming to render … Read the Rest »
Meet the Mad Scientist Who Helped Create Kelly Slater’s Surf Ranch
A French engineer, a circular-pool dream and thousands of simulations collide in the untold story behind Kelly Slater’s Surf Ranch and the next generation of wave pools.
Scientists created one of the largest simulations of our universe ever — about the size of 500,000 HD movies
These giant "virtual universe" simulations help scientists trace cosmic evolution across billions of years.
Practicing today for tomorrow’s emergencies – WHO convenes countries and partners to simulate response to major disease outbreak
WHO wrapped up Exercise Polaris II, a 2-day high-level simulation exercise, based around an outbreak of a fictional new bacterium spreading across the world. Bringing together 26 countries and territories, 600 health emergency experts and over 25 partners, the exercise allowed countries to test their preparedness for pandemics and other major health emergencies.
Texas Health brings emergency medical simulation training to Johnson County EMS
Texas Health Fort Worth began deploying the new mobile education unit in January.
A bizarre new state of matter may be hiding inside Uranus and Neptune
Deep inside planets like Uranus and Neptune, scientists may have uncovered a bizarre new state of matter where atoms behave in unexpected ways. Advanced simulations suggest that carbon and hydrogen, under crushing pressures and scorching temperatures, can form a strange hybrid phase—part solid, part fluid—where hydrogen atoms spiral through a rigid carbon framework. This unusual “superionic” structure could reshape how heat and electricity flow inside these distant worlds, potentially helping explain their mysterious magnetic fields.
Special Seminar
Speaker – Lenny Smith, Professor, Virginia Tech. Hosted by Walt Robinson. Title – Reflection versus Regurgitation: Simulation, Statistics, and Insight from Fitzroy to ECMWF’s AIFS. (link to relevant paper) Abstract – There is considerable…
Teaching Legislative Politics Through a Game: Active Learning, Assignment Co-creation, and Assessment Across Modalities -
Teaching Legislative Politics Through a Game: Active Learning, Assignment Co-creation, and Assessment Across Modalities By Jennifer L. Clemens, University of Wisconsin–Parkside and Michael A. Hansen, University of Turku This article examines a legislative politics simulation [...]
NBA playoff simulation: Thunder favored, but will they win it all?
The defending champion Oklahoma City Thunder are good bets to repeat, but we ran 100 different simulations to find out what other teams could pull the upset.
Saturn’s largest moon could see 10-foot waves from a tiny breeze
Surf's up, Titan.
How low should blood pressure go? Science has the answer
New research suggests that aiming for a lower blood pressure target may deliver bigger heart health benefits than previously thought. Using large datasets and simulation models, scientists found that keeping systolic blood pressure below 120 mm Hg could reduce the risk of heart attack, stroke, and heart failure more than higher targets.
ICARUS experiment marks major milestone in first neutrino science results
The milestone by ICARUS paves the way to probe sterile‑neutrino oscillations for the full Short-Baseline Neutrino Program at Fermilab. The first results demonstrate the data’s excellent quality and suitability for physics analyses, as well as the maturity of the software tools for event selection, fitting and detector simulation.
Building Custom Atomistic Simulation Workflows for Chemistry and Materials Science with NVIDIA ALCHEMI Toolkit
For decades, computational chemistry has faced a tug-of-war between accuracy and speed. Ab initio methods like density functional theory (DFT) provide high fidelity but are computationally expensive…
Scientists were wrong about lifespan. Your genes matter way more than we thought
For years, scientists believed our lifespan was mostly shaped by environment and chance, with genetics playing only a minor role. But a new study from the Weizmann Institute flips that idea on its head, revealing that genes may actually account for about half of the differences in how long people live. By analyzing massive twin datasets—including twins raised apart—and using innovative simulations to filter out deaths from accidents and other external causes, researchers uncovered a hidden genetic influence that had been masked for decades.
“AI polls” are fake polls
A few weeks after Donald Trump’s second presidential win, I took the train up from London (where I was living at the time) to Oxford to attend a conference on polls and forecasts of the 2024 election. Most of the attendees were pollsters or academics, but I also watched presentations from Aaru and Electric Twin, two companies that do what is interchangeably called synthetic sampling, silicon sampling, or synthetic audiences. Stripped of startup jargon, that means they use large language models (LLMs) to simulate responses to public opinion polls by having AI agents take on the role of survey respondents. I had already heard of Aaru thanks to some articles with eye-catching headlines like “No people, no problem: AI chatbots predict elections better than humans” in the months leading up to Election Day. The founders were making some big, some might even say far-fetched claims, such as: “within two years, we will simulate the entire globe — from the way crops are grown in Ukraine to how that impacts production of oil in Iraq, trade through the strait of Malacca, and elections for the mayor of Baltimore.” When Semafor asked Aaru’s cofounders — Cameron Fink and Ned Koh — about my boss, they said “we respect all those who came before us.” Nate (as he so often does) shared his thoughts on Twitter: Fink and Koh were relatively good-natured about this back-and-forth when we spoke at Oxford. They even offered to mail me one of the t-shirts featuring Nate’s quote they apparently had made. I never took them up on the offer, which I now somewhat regret. These synthetic sampling companies fell off my radar for a while, but they do still exist. In fact, Aaru recently received a $1 billion valuation. Is what they’re doing anywhere close to the most important frontier in AI development? Not by a longshot, especially when Anthropic just developed a model so adept at exploiting software vulnerabilities that it’s only being released to 40 companies. Still, silicon sampling is increasingly finding its way into public polling. Axios reported in March that “a majority of people trust their own doctors and nurses” based on findings from Aaru — without mentioning that the “people” in that sentence were actually LLMs. Around the same time, the Public Sentiment Institute “boosted” their online sample of 373 real survey respondents with 114 AI agents.1 (Spoiler alert: even the co-founder of Electric Twin doesn’t think that’s a particularly defensible approach.) Polling companies like Qualtrics and Ipsos are also developing synthetic data panels. So, what should we make of these … “polls”? Let’s get one thing out of the way: whatever they are, they’re not polls in the way that term is usually defined. Subscribe You can’t replace polls with AI On one hand, using LLMs to essentially make up fake survey respondents sort of sounds like the dumbest idea ever, one that will at best imperfectly replicate real polls while introducing all sorts of biases. On the other hand, with LLMs improving at a remarkable, perhaps even alarming rate, maybe that means I’m a dinosaur at the ripe old age of 24 because I still want to rely on polls that talk to actual people. I’m not going to argue that synthetic samples are completely useless. In fact, as I’ll return to later, there is evidence that some techniques can replicate topline survey results quickly and cheaply. But the marketing from certain companies can be slightly optimistic. “No traditional poll will exist by the time the next general election occurs,” said Fink in 2024. We’re just 206 days away from the midterms, and based on the fact that I still have to collect a bunch of polls every day, I’d say he should have run that prediction by a sample of AI agents before the interview.2 To see why synthetic samples can’t replace polls, here’s a quick primer on how they work. The simplest version of these models involves taking a LLM (like ChatGPT or Claude), giving it a demographic profile (e.g., a white, college-educated woman who lives in Utah and makes $70k a year), and then asking it to respond to a survey question. You repeat that process a few thousand times using different demographic profiles and end up with a sample of synthetic survey responses. The actual models used by private companies are more sophisticated than this, usually because they incorporate more hypothesized demographic characteristics for each agent and provide them with extra information. Aaru, for example, feeds agents a diet of news and information they’d be likely to consume, while Electric Twin incorporates their customers’ proprietary data about the audience they’re trying to replicate. The way Ben Warner, the co-founder of Electric Twin, explained it to me was “we have a large amount of data on […] for instance, 5,000 people. Can we make an accurate prediction of how they would respond to another question?” Still, it should be obvious why synthetic samples can’t replace polls. Polling is fundamentally a data collection process. We might use surveys to make predictions by feeding them into election forecasts, but the main purpose of a poll isn’t prediction, it’s gathering new data about what people think and how they feel. Silicon sampling, on the other hand, produces no new data. It’s simply a model: you input LLM training data, demographic prompts, and a bunch of other information, and it spits out a prediction for what a poll would say. We love models here, but models aren’t polls. That difference is an important philosophical sticking point for most pollsters I talk to. “I think politics should stay away from [synthetic sampling], because we’re trying to […] represent the voice of the people,” said Natalie Jackson, a vice president at GQR Insights. Democratic pollster John Hagner told me: “I think I’m just incredibly skeptical of this idea. I don’t think it’s research. At that point, you’re asking the machine to tell you what you already believe.” Hagner has seen some presentations of early synthetic sampling experiments, but so far, “if it’s being used in a campaign, people are keeping it incredibly quiet.”3 But Eli, I hear you saying, aren’t polls themselves increasingly governed by modeling decisions? Indeed they are: pollsters’ choices on which sampling method to use, how to define their likely voter models, and how to weight their samples can and do lead to dramatic differences in the results they publish. Aaru even referenced these limitations in the methodology statement included with that maternal mortality “poll” — although I’m using the term “methodology statement” loosely here, because it doesn’t really explain how the model works at all. We can ignore the (frankly preposterous) implication that synthetic sampling isn’t subject to a separate set of biases. The important point is that there’s still a meaningful difference between using weighting and other statistical techniques on actual polling data and using a model to predict what a poll would say. The latter is far closer to election forecasts or techniques like MRP — potentially useful models, but not a replacement for polls.4 To be fair, other synthetic sampling companies are perfectly happy with the distinction between polls and models. Warner compared polling and synthetic sampling to different tools in a toolbox. “The mistake I think we make is we think that these new tools should either work in exactly the same way or somehow replace these old tools,” he said. “Rather than thinking of it as, okay, so we’ve always had the hammer, we’ve always had the screwdriver, now we’ve got a saw. But don’t use a saw to try [to] do the job of a hammer.” A quick comment from Nate Eli didn’t ask me for a comment — rather rude of him, don’t you think? But since I’m editing this story, I figured I’d add a few quick thoughts rather than putting words in his mouth. Beyond the frequently misleading marketing, what bothers me about the AI “poll” hype is that as AI tools make statistical inference cheaper and/or better (note that these are not synonyms) that actually increases the comparative value of collecting original data. You might be able to train a model to make a reasonable estimate of what some hard-to-reach poll respondent would say — say, a young Black man who voted for Trump. (Such a person checks a number of boxes for a voter who is usually hard to reach in surveys.) Indeed, this is closely related to what models like the Silver Bulletin forecast already do. They essentially smooth out the kinks in noisy survey data by making inferences based on past voting patterns or national polls or surveys of other states. But you don’t actually know what these voters think unless you’re reaching them directly. If there’s a shift in opinion among this subgroup, you’re not going to detect it. So if I were running a campaign, I’d invest more in going the extra mile to find a representative sample of those voters. And then I’d hire some smart quants — or Claude? — to figure out the implications for campaign strategy based on proprietary data that my competitors didn’t have access to. -Nate Silver Are these models any good? If synthetic surveys are just a new type of model, the next obvious question is whether the models are at least accurate. The answer very much depends on who you ask. On one end of the spectrum, you have the maximalist argument that synthetic sampling is more accurate than actual polls. “It’s an incredibly challenging problem to go to someone and say ‘hey, we’re going to be more accurate at predicting human behavior than you, even when you talk to your customers directly’,” Koh recently told CNBC. In his view, synthetic sampling isn’t a saw to polling’s hammer, it’s “magic.” There’s certainly evidence that synthetic samples can replicate certain survey toplines. But if Aaru does have any examples of their approach outperforming the polls, they’re keeping those to themselves.5 Aaru’s 2024 election model, for example, had Kamala Harris leading in Michigan, Nevada, Pennsylvania, and Wisconsin on November 4th. And although they’ve since taken down their forecast page, they gave Harris a 50.5 percent chance of winning the race on November 2nd.6 After the election, Fink told Semafor he was happy enough with those results because they were “within margin of error,” a term that is completely meaningless when applied to a “sample” of AI agents. And of course, Aaru says their models have improved since 2024, so supposedly now they’d be more accurate than the polls? Still, their stronger argument is on cost: “We are significantly faster and cheaper than traditional polling, and still more accurate,” said Fink. The first two claims are undeniably true, but the third brings us to the opposite end of the spectrum. Both Jackson and Hagner are skeptical that these models are reliable for anything beyond replicating common survey toplines. “I just […] don’t think the machines are what we want when we’re looking for nuanced views. My example on this is people in Arizona and Nevada in 2024 who voted for Trump and voted for expanding abortion in their states on ballot initiatives,” said Jackson. Hagner identified another issue. Maybe the synthetic respondents, like sycophantic LLMs, are inhuman in one important way: they’re too nice. “The reports that have come through at the meetings that I’ve been at are that the early experiments on this, they cannot get respondents to be as racist or sexist or, frankly, as negative as human respondents,” he said. Academic research mostly agrees on this point. While there are some papers that show promising results when using LLMs to replicate polling data, most show that LLMs suffer from various quirks like producing too few “don’t know” responses and can seriously overpredict the favorability of politicians like Donald Trump and Kamala Harris. They also seem to struggle with too little variation between demographic subgroups, so the difference in predicted opinion between Democrats and Republicans, for example, is too small. When I asked Warner about these studies, his response to these papers was that just because academics can’t get synthetic sampling to work doesn’t mean that the technique doesn’t work in general. “Actually, the argument is, okay, yours does not [work]. That does not mean […] for this complex set of machinery, which uses a lot of investment, a lot of time, a lot of money, you can’t get it to work.” Cards on the table, I’m somewhat sympathetic to this argument because academics aren’t exactly great at making election forecasts. Usually, the people with skin in the game are the most accurate. Warner’s argument is that the approach Electric Twin takes — which includes, for example, making multiple predictions for each synthetic respondent using different models and prompts and subsequently averaging those to get a final prediction in a sort of ensemble forecast — produces better results than the simpler academic models. Warner shared a comparison between his method and the method from a recent academic paper with me, and Electric Twin was indeed able to get more accurate replication. But even still, he acknowledged that synthetic sampling “is not a crystal ball.” “If you asked me, do I think using other data sources will be more accurate than asking somebody who they will vote for, I would probably say no. But if you asked me ‘would your system be useful for our turnout modeling today?’ I would say yes.” For better or worse, it looks like the method is already getting more popular in the market research world. Most of the clients Aaru touts these days are businesses like EY and McDonald’s. And AI will probably begin to pop up in other parts of the political polling process. Pollsters are already using it to code open-ended survey responses, and some firms, like YouGov, are testing using LLMs to ask survey respondents questions. More worryingly, one danger to actual polls is that AI agents can be used to infiltrate online surveys. Most online polls use various checks to prevent that from happening, but there’s conflicting evidence on how effective those filters are and how prevalent AI agents currently are in online panels. If those agents ever become impossible to detect, it might spell the end of online polling, but the solution isn’t to replace all of your respondents with ChatGPT. Silver Bulletin is a reader-supported publication. To receive new posts and support our work, consider becoming a subscriber. Subscribe 1 That particular poll obviously doesn’t meet Silver Bulletin standards for aggregation. But we exclude all Public Sentiment Institute polls from our averages because we classify them as an amateur polling firm. 2 You could argue that Fink meant the next presidential election, but (a) I’m also confident we’ll still have real polls in 2028 and (b) in that case he should have asked an LLM to define “general election.” 3 Quick caveat: that’s reporting from a Democratic pollster. It’s possible that Republicans are more willing to use AI in political campaigns. 4 Indeed, Silver Bulletin does not include “polls” produced by MRP in our forecasts or averages, and we think it’s extremely misleading when their practitioners describe them in a way that suggests original data had been collected among a large number of states or Congressional districts. 5 A recent report from Aaru and EY did show two examples of a synthetic estimate being closer than a survey to a real-world benchmark — but I’d take those findings with a grain of salt because the report reads more like an ad and didn’t involve any sort of prediction being made ahead of time. 6 For comparison, our odds for Harris on the same day were 48.2 percent.
If Venus Has Life It May Have Come From Earth, Scientists Say
Could it be?
Gemini gains the ability to create interactive simulations
And you can change their parameters on the fly. The latest Gemini feature lets Google's AI generate interactive simulations and models. With this, Gemini...
Embedded mental health clinical interns quickly connect people to support
Counseling Service of Addison County’s embedded master’s-level clinical internship program connects people to services, offering support from a familiar face within a trusted agency. Interns gain close supervision and guidance from expert staff in an established Community Mental Health Agency.
Institute of Industrial Science, The University of Tokyo and George Mason University scientists mine past air temperature forecasts, lower cost with greater subseasonal prediction accuracy
Researchers at Institute of Industrial Science, The University of Tokyo and George Mason University‘s College of Science have developed a new method that improves air temperature forecasts one to five weeks in advance—without requiring additional model simulations.
Physicists just solved a strange fusion mystery that stumped experts
Fusion scientists have solved a long-standing mystery inside tokamaks, the donut-shaped machines designed to harness fusion energy. For years, experiments showed that escaping plasma particles hit one side of the exhaust system far more than the other, but simulations couldn’t explain why. Now, researchers have discovered that the rotation of the plasma itself plays a crucial role—working together with sideways particle drift to create the imbalance.
Some dinosaurs could rise up like giants — until they grew too big
Certain smaller sauropods could stand on their hind legs with surprising ease, giving them access to higher food and a defensive edge. Computer simulations show their bones handled stress better than those of their larger relatives. However, as they grew, the sheer weight made this posture much harder to sustain. What started as a useful trick in youth became a more limited, strategic move in adulthood.
NFL mock draft: Another 7-round Arizona Cardinals simulation
In this seven-round simulation, the Cardinals make one trade and land players to fill many needs.
Southwest Virginia Higher Education Center opens regional health training simulation lab
The Southwest Virginia Higher Education Center hosted a ribbon-cutting for its new regional simulation lab.
Photos: Pro Football Cares Sports Emergency Response Training at Raiders HQ
A Pro Football Cares Sports Emergency Response Training was held at Raiders HQ where athletic trainers and coaches were able to participate in a series of simulations on football equipment removal, cardiac arrest and exertional heat stroke.
Supercomputers just solved a 50-year-old mystery about giant stars
Astronomers have finally cracked a decades-old mystery about red giant stars—how material from their deep interiors makes its way to the surface. Using cutting-edge supercomputer simulations, researchers discovered that stellar rotation plays a powerful role in mixing elements across a previously unexplained barrier inside the star.
Spring Health Fair delivers commitment to student wellness
The Student Wellness Advisory Committee hosted their annual Spring Health Fair on Friday afternoon where students were able to gain free tote bags, learn from live simulations and gain insight on new clubs and organizations centered on wellness.
Full 7-round Arizona Cardinals mock draft simulation
We have a new 7-round mock draft for the Arizona Cardinals.
Cracking the human simulation code: Aaru co-founders on refining the science of prediction
Cameron Fink, Aaru co-founder and CEO, Ned Koh, Aaru co-founder and president, and John Kessler, Aaru co-founder and CTO, join 'Squawk Box' to discuss the origins of the company, using AI to predict human behavior with higher accuracy and speed than traditional methods, and more.
NCAA Tournament 2026 bracket: Simulation predicts surprising upsets, sleepers, March Madness picks
When does the March Madness 2026 schedule start, and when are 2026 NCAA Tournament brackets due? The 2026 March Madness bracket begins on Thursday, March 19
CBA Launches Small Business Curriculum as Part of Advanced Education Programs
WASHINGTON, D.C. – The Consumer Bankers Association (CBA) recently launched the newest addition to its banking education curriculum, Talking Business with Small Business (TBWSB) – a curriculum focused on small business lending – as part of its Advanced Education programs. The curriculum leverages a tabletop simulation for branch managers to better understand…
NCAA Tournament 2026 bracket: Top-rated simulation predicts surprising upsets, March Madness picks, sleepers
When does the March Madness 2026 schedule start, and when are 2026 NCAA Tournament brackets due? The 2026 March Madness bracket begins on Thursday, March 19
Strengthening One Health Field Epidemiology Capacity in the WHO South-East Asia Region
Recent public health emergencies—including the COVID-19 pandemic, recurrent zoonotic disease outbreaks, and the high level of antimicrobial resistance—have underscored the urgent need for strong, multisectoral field epidemiology capacities. This reinforces the importance of a workforce equipped with One Health field epidemiology competencies. To address this need, WHO works with the Food and Agriculture Organization of the United Nations (FAO), the United Nations Environment Programme (UNEP), and the World Organisation for Animal Health (WOAH), as part of Quadripartite Alliance for One Health, to strengthen field epidemiology training programmes using a One Health approach.These efforts are aligned with WHO SEARO’s Strategic Framework for Action for Strengthening Surveillance, Risk Assessment and Field Epidemiology and the Regional Roadmap to Advance Field Epidemiology Capacities 2025–2029 emphasize workforce development, multisectoral collaboration, and enhanced preparedness for health emergencies. With funding from the Pandemic Fund, The Competency for One Health Field Epidemiology (COHFE) integration within Field Epidemiology Training Programmes (FETPs) or similar initiatives has been initiated in Bangladesh and India and is planned in Sri Lanka.BangladeshIn Bangladesh, COHFE is being adapted to strengthen the Field Epidemiology Training Programme, Bangladesh (FETP-B) Advanced programme. Initiated in July 2025 and led by the Institute of Epidemiology, Disease Control and Research (IEDCR), the process brought together stakeholders from human, animal, and environmental health sectors, academia, and laboratories.A Multisectoral Working Group (MWG) was established to guide implementation, conduct readiness assessments, and identify opportunities to institutionalize One Health collaboration. Through national multisectoral consultations and a competency prioritization workshop held in Dhaka in November 2025, stakeholders identified priority technical competencies for surveillance, outbreak investigation, data analysis, and risk assessment, alongside functional competencies related to leadership, coordination, and the laboratory–epidemiology interface. The process also surfaced systemic barriers—such as challenges in data sharing and limited joint investigations—enabling targeted recommendations. These outcomes are guiding curriculum revision, the introduction of One Health simulations and joint field exercises, and the development of a national multisectoral mentorship pool, with updated training materials expected to be implemented in the upcoming cohort.Figure 1. Competency prioritization involving multisectoral stakeholders“COHFE represents a significant step forward in operationalizing the One Health approach. The collaboration between national institutions and the Quadripartite partners is critical for building resilient systems to address complex health challenges in the region.”, Prof. (Dr) Tahmina Sirin, Director, Institute of Epidemiology, Disease Control and Research (IEDCR), Bangladesh.Aligning India’s One Health Training with COHFEIn India, the National Centre for Disease Control (NCDC) has strengthened One Health workforce development through its SectorConnect initiative. WHO supported a comprehensive competency mapping and curriculum review to align the Field Epidemiology Programme in One Health (FEP-OH) with all 14 COHFE domains. The review identified strong foundations while generating actionable recommendations to strengthen One Health learning outcomes, expand joint field-based learning, and enhance competency-based assessment. These improvements are informing programme refinement and phased scale-up across additional states.Figure 2. Strengthening learning outcomes for the SectorConnect FEP OH CurriculumThrough these efforts, WHO SEARO will continue to support countries to move from separate sector-based training toward an integrated, impact-driven One Health field epidemiology workforce capable of addressing complex health threats at human-animal-environment interface.ReferencesFood and Agriculture Organization of the United Nations, World Health Organization, & World Organisation for Animal Health. (2024). Competencies for One Health field epidemiology (COHFE) framework. https://www.who.int/publications/i/item/9789240080058World Health Organization. (2023). Strategic framework for action for strengthening surveillance, risk assessment and field epidemiology for health security threats in the WHO South-East Asia Region. WHO Regional Office for South-East Asia. https://www.who.int/publications/i/item/9789290210030World Health Organization. (2024). Regional roadmap to advance field epidemiology capacities in the WHO South-East Asia Region 2025–2029. WHO Regional Office for South-East Asia. https://www.who.int/publications/i/item/9789290229544
THOR AI solves a 100-year-old physics problem in seconds
A new AI framework called THOR is transforming how scientists calculate the behavior of atoms inside materials. Instead of relying on slow simulations that take weeks of supercomputer time, the system uses tensor network mathematics and machine-learning models to solve the problem directly. The approach can compute key thermodynamic properties hundreds of times faster while preserving accuracy. Researchers say this could accelerate discoveries in materials science, physics, and chemistry.
'Stealth hit' Pokemon game sends Nintendo shares soaring
Fan buzz around life-simulation game "Pokemon Pokopia" sent Nintendo shares soaring on Wednesday, with some hailing the new title as a welcome antidote to global conflicts.Nintendo shares were up nine percent in mid-morning trade on Wednesday, also likely boosted by the release of the final trailer for the star-studded upcoming "Super Mario" movie sequel.
Simulation Training Featured in New Episode of NYC Health + Hospitals Podcast The Remedy - NYC Health + Hospitals
New York, NY – NYC Health + Hospitals today released a new episode of its podcast, The Remedy, featuring leaders and health care providers from the
Conway Regional opens simulation lab to train future healthcare workers
Conway Regional Health System unveiled a new simulation lab Thursday designed to train the next generation of healthcare workers and help address workforce shor
Astronomers discover giant cosmic sheet around the Milky Way
For decades, astronomers wondered why most nearby galaxies are speeding away from the Milky Way instead of being pulled in by its gravity. New simulations reveal the answer: our galaxy sits in a gigantic, flat sheet of matter surrounded by huge empty voids. This hidden structure—dominated by dark matter—balances gravitational forces and lets neighboring galaxies drift outward. The discovery finally explains the puzzling motions of galaxies just beyond our Local Group.
February 27, 2026
A new simulation shows large amounts of hydrogen in our planet’s core. Scientists report on what they found in 290-million-year-old vomit. Plus, a comedic play and a 20-year neurology study explore what we can do to prevent dementia and cognitive decline. And, a federal law aims to close the legal loophole that lets stores sell THC products from hemp.
PRET for impact: Advancing pandemic preparedness in the Democratic People’s Republic of Korea
IN THE SPOTLIGHT PRET for impact: Advancing pandemic preparedness in the Democratic People’s Republic of KoreaThe Democratic People’s Republic of Korea (DPR Korea) is making significant progress in strengthening pandemic preparedness, with dedicated support from the WHO South‑East Asia Regional Office (SEARO) and the Pandemic Influenza Preparedness (PIP) Partnership Contribution (PC).Aligned with the WHO Global Influenza Strategy and the Pandemic Influenza Preparedness (PIP) Framework Partnership Contribution High‑Level Implementation Plan III (2024–2030), the country is strengthening its capacity to prepare for and respond to influenza and other respiratory threats. In early 2025, a core team from the Ministry of Public Health of the Democratic People’s Republic of Korea (DPRK) participated in a two‑day virtual workshop, which included a mini‑simulation exercise. The workshop was organized by WHO SEARO in collaboration with the Preparedness and Resilience for Emerging Threats (PRET) Secretariat at WHO headquarters and the WHO Country Office for the DPRK. Read more A REGIONAL CONSULTATION GUIDES WHO SEARO IN DEVELOPING A ROADMAP FOR IMPLEMENTING THE PREPPAREDNESS AND RESILIENCE FOR EMERGING THREATS (PRET) INITIATIVE BASED RESPIRATORY PATHOGEN PANDEMIC PREPAREDNESSInfluenza and other respiratory pathogens with pandemic potential continue to pose major public health risks globally. To strengthen regional preparedness and response, Member States of the WHO South‑East Asia Region (SEAR) convened in Colombo, Sri Lanka, to review and provide input on the draft Roadmap and Action Framework for Broad Respiratory Pathogen Pandemic Planning (2026–2031). Supported through the Pandemic Influenza Preparedness (PIP) Partnership Contribution (PC) funds, the proposed roadmap aims to advance robust, integrated preparedness and response capacities for influenza and other respiratory threats across SEAR. This work is aligned with the Preparedness and Resilience for Emerging Threats (PRET) initiative and the SEAR Strategic Roadmap on Health Security and Health System Resilience for Emergencies Read moreSRI LANKA’S USE OF ANTIVIRALS TO REDUCE THE SEASONAL INFLUENZA BURDEN Sri Lanka is strengthening its influenza response by placing greater emphasis on rapid testing, timely treatment, and integration with maternal health services. With support from WHO, the country is building a stronger evidence base on the burden of influenza and the value of vaccination, advancing a holistic approach to protect vulnerable populations and enhance pandemic preparedness. To reduce hospitalizations and deaths from seasonal influenza, Sri Lanka is taking proactive steps to expand access to rapid diagnostic testing and ensure prompt initiation of treatment. National clinical guidelines recommend the use of oseltamivir for individuals at high risk of complications, helping prevent severe illness and shorten the duration of infection. This strategy not only saves lives but also supports local pharmaceutical production, contributing to a reliable supply of antivirals for both seasonal outbreaks and future pandemics.Read moreBHUTAN BOOSTS PANDEMIC INFLUENZA READINESS WITH PANDEMIC INFLUENZA PREPAREDNESS FRAMEWORK PARTNERSHIP CONTRIBUTION SUPPORTED TRAINING AND SIMULATIONSOriginally developed in 2012, the National Influenza Pandemic Preparedness and Response Plan (NIPPRP) has long served as Bhutan’s central framework for managing influenza threats. In 2024, the plan underwent a comprehensive update to reflect evolving global and regional guidance, lessons learned from the COVID‑19 pandemic, and new capacities developed under the PIP Framework. The revised NIPPRP is now more clearly aligned with Bhutan’s broader emergency and disaster risk management structures, ensuring a coordinated, whole‑of‑government approach to pandemic response. As part of efforts to strengthen national pandemic preparedness, Bhutan conducted three national‑level training sessions followed by simulation exercises on influenza pandemic preparedness and response at five sentinel hospitals between April and May 2025, with support from the PIP Partnership Contribution (PC). These activities brought together more than 140 health professionals and key partners from across the country, with a focus on operationalizing the updated NIPPRP and enhancing readiness at key frontline facilities. Read moreSTRENGTHENING INFLUENZA SURVEILLANCE: WHO BRINGS TOGETHER INFLUENZA EXPERTS FROM SOUTH-EAST ASIA AND WESTERN PACIFIC REGIONSThe World Health Organization (WHO) convened the 18th Bi‑regional Meeting of National Influenza Centres (NICs) and Influenza Surveillance for the WHO South‑East Asia and Western Pacific Regions from 9–11 September 2025. Hosted by the WHO South‑East Asia Regional Office (SEARO), the meeting brought together representatives from Member States, Territories, and Areas across both regions, alongside experts from the WHO Collaborating Centres for Reference and Research on Influenza in Australia and Japan, the WHO Collaborating Centre for Studies on the Ecology of Influenza in Animals in the United States, quadripartite partners (WOAH and FAO), and other technical counterparts. The consultation provided a platform to review progress, share experiences, and outline priorities for strengthening influenza and respiratory pathogen surveillance across the two WHO regions.Read moreNETWORKS AND TECHNICAL PARTNERSHIPSSTRENGTHENING PANDEMIC READINES: SOUTH-EAST ASIA ADVANCES UNITY STUDIES UNDER THE PANDEMIC INFLUENZA PREPAREDNESS INITIATIVEThe WHO Global Influenza Surveillance and Response System (GISRS) Investigations & Studies initiative, also known as Unity Studies, provides a standardized, globally coordinated framework to support these efforts. Originally developed for influenza and adapted during COVID-19, Unity Studies enable countries to conduct rapid, nationally led investigations and studies to assess transmissibility, severity, susceptibility, and intervention effectiveness. To better prepare for future influenza pandemics, WHO has recently launched a network of sites which will be ready to conduct these investigations and studies rapidly in the event of a pandemic, or epidemic of an emerging or re-emerging respiratory pathogen. The WHO South-East Asia Region now hosts six sites of WHO GISRS Unity Studies across the five countries including Sri Lanka, strengthening regional capacity for early pandemic investigations. From 13–16 July 2025 in Colombo, the Global Public Health Summit offered an opportunity to advance this agenda. Read moreSTRENGTHENING PANDEMIC PREPAREDNESS THROUGH UNITY STUDIES: REAL-WORLD APPLICATIONS FROM INDIA IN WHO’S SOUTH-EAST ASIA REGION As countries continue to strengthen their pandemic preparedness through investments in resilient health systems during the interpandemic period, the World Health Organization’s (WHO) Global Influenza Surveillance and Response System (GISRS) Investigations & Studies Global Initiative, Global Initiative, commonly known as Unity Studies, remains a vital framework for respiratory pathogen pandemic preparedness. These standardized epidemiological protocols enable countries to rapidly generate high‑quality data on transmissibility, infection severity, population immunity, and the effectiveness of public health interventions. To highlight how Unity Studies protocols are being adapted and applied in national settings, WHO Headquarters, together with WHO Regional Offices, including the WHO South‑East Asia Regional Office (SEARO), hosted a global webinar. The session showcased case studies from India and the United Kingdom (UK), demonstrating innovative approaches to respiratory virus surveillance and pandemic readiness. Read moreSTRENGTHENING REGIONAL INFLUENZA SURVEILLANCE AND PREPAREDNESS: WHO SEARO HOSTS HIGH-LEVEL WEBINAR AMID RISING H5 THREATS In response to the increasing detection of avian influenza A(H5) cases in humans globally and within the WHO South‑East Asia Region, the WHO Regional Office for South‑East Asia (SEARO) convened a regional webinar on 25 July 2025 to review preparedness mechanisms, strengthen surveillance pathways, and promote timely reporting of unusual respiratory events. The virtual session brought together more than 50 participants from Ministries of Health, National Influenza Centres (NICs), National IHR Focal Points (NFPs), WHO Headquarters, WHO Country Offices across the Region, and WHO Collaborating Centres (CCs) for influenza. The webinar provided an opportunity to share situational updates, discuss surveillance gaps, and reinforce coordination mechanisms critical for early detection and rapid response to avian influenza threats. Read moreOther Key PIP Updates (July-December 2025)MEMBER STATES:Bangladesh drafted and conducted a series of Core Group Meetings in 2025 to review and update the National Pandemic Preparedness and Response Plan for Respiratory Pathogens. In addition, the country carried out Rapid Response Team (RRT) trainings on seasonal and avian influenza and other re‑emerging infectious diseases. Bangladesh also updated its epidemiological data‑training programme for physicians at sentinel surveillance sites, using a revised epidemiological data‑collection tool.Bhutan revised and updated its National Influenza Pandemic Preparedness and Response Plan (NIPPRP) in 2025 and developed a national Standard Operating Procedure (SOP) for influenza outbreak management. DPR Korea conducted a nationwide simulation exercise in July 2025 to support the update of its National Influenza Pandemic Preparedness Plan (2019). This followed a PRET‑based virtual orientation and mini‑simulation exercise held in May 2025 by SEARO and the WHO PRET Secretariat. The comprehensive exercise, conducted across central and provincial levels, assessed strengths and gaps in pandemic influenza preparedness and response. The feedback gathered has been incorporated into the revised draft plan to address identified weaknesses and strengthen overall readiness.India conducted a targeted nationwide online re‑orientation training on review and updates to the influenza portal for all State Surveillance Officers (SSOs) and relevant officials using the Integrated Health Information Platform (IHIP). The training emphasized correct marking and reporting of influenza samples as “unsubtypable” or “not subtyped” where applicable, to ensure accurate national surveillance.Maldives conducted simulation exercises in three atolls based on its existing National Pandemic Preparedness and Response Plan (NPPRP). The country plans to conduct similar exercises in three additional atolls after updating the NPPRP in 2026. In addition, Maldives conducted training on FluID and FluMart as part of a broader data‑management training package for Ministry of Health and Health Protection Agency officials, aimed at improving routine reporting to WHO platforms and the DHIS2 system.In Myanmar, technical support was provided to the National Influenza Centre (NIC) through dedicated human‑resource support (January–December 2025) to sustain data management and laboratory detection capacity for influenza and other emerging and re‑emerging respiratory viruses. This support also included strengthening genomic sequencing capacity.Nepal conducted a consultative workshop on the National Pandemic Preparedness and Response Plan (NPPRP) in July 2025. The workshop engaged key stakeholders from multiple sectors to review lessons from past pandemic responses, assess current mechanisms, and gather critical inputs for strengthening national preparedness. Key recommendations included revising the 2019 NPPRP and developing an updated Influenza Pandemic Preparedness Plan aligned with emerging risks, national priorities, and relevant international frameworks. Nepal also held a two day workshop dedicated to advancing the National Action Plan for Genomic Surveillance, aimed at reinforcing the country’s capacities for early detection and monitoring of emerging respiratory pathogens.Sri Lanka received the final report from WHO on the feasibility assessment for conducting an Influenza Burden of Disease (BoD) study in the country. The assessment, conducted from 28 April to 9 May 2025, included a desk review of existing data sources, document review, visits to sentinel surveillance sites, and key informant interviews. The study identified critical data sources required for a future BoD assessment, as well as their availability and quality. Officials from the Epidemiology Unit, the National Influenza Centre (NIC), and SARI/ILI sentinel hospitals across the country participated in the assessment.Timor Leste conducted on‑the‑job training and monitoring activities at two sentinel surveillance sites to strengthen surveillance and laboratory capacity for the detection of SARI, ILI, ARI, and other respiratory infections. REGIONAL OFFICE:SEARO conducted a Member States Consultation on the first draft of the Regional Roadmap and Strategic Action Framework for implementing WHO’s PRET‑based broad respiratory pathogen pandemic planning for 2026–2031. The consultation took place on 29–30 October 2025 during the final two days of the Regional PIP PC Annual Review and PRET Meeting in Colombo, Sri Lanka. Participants included representatives from Member States, WHO Country Offices, WHO SEARO, WHO Headquarters, partner agencies, and invited multidisciplinary technical experts. They reviewed the draft and provided detailed feedback, which has since been incorporated into a revised version. The consultation helped shape a shared regional vision for robust, PRET‑aligned pandemic preparedness, building on country‑level progress, addressing existing gaps and challenges, and aligning with global and regional health security frameworks. The publication is planned for release during the SEAR regional committee in 2026.SEARO documented the evolution of the National Influenza Pandemic Preparedness Planning process in the Region, tracing its development from its origins as a planning tool for responding to avian influenza outbreaks in 2006, through the experiences of the 2009 and 2019 pandemics, to the current status. The document outlines the way forward for respiratory pathogen pandemic planning across all 10 SEAR countries, aligning future national plans with WHO’s PRET Framework for broad respiratory pathogen pandemic preparedness.The Infectious Hazard Management (IHM) Unit has drafted the WHO South‑East Asia Regional Guidance on Influenza Prevention and Control, aligned with the Global Influenza Strategy and other regional initiatives. The draft is currently under review. Although this technical product was developed in response to a request from the Ministry of Public Health of the Democratic People’s Republic of Korea (MoPH‑DPRK), SEARO is finalizing it for use not only in DPRK but also across all Member States in the Region.The Immunization and Vaccine Development (IVD) Unit, in collaboration with WHE/SEARO, completed a regional landscape analysis on key barriers, challenges, and opportunities for the introduction of seasonal influenza vaccination in WHO South‑East Asia Regional (SEAR) countries.Additionally, the Immunization and Vaccine Development (IVD) Unit has completed a regional landscape analysis of influenza vaccine procurement systems across the WHO South‑East Asia Region (SEAR).KEY MEETINGS AND PUBLICATIONSMEETINGS:Six-month progress review of the implementation of the PIP-PC workplans in SEAR, focusing on Outputs 1 and 2 for the period January to June 2025, and held virtually across three sessions on 4, 8, and 10 July 2025. Read moreAnnual Regional Meeting (Hybrid) on Pandemic Influenza Preparedness (PIP) Partnership Contribution (PC) and WHO’s Preparedness and Resilience for Emerging Threats (PRET) Implementation in SEAR in Colombo, Sri Lanka, 27 to 30 October 2025. Read more18th Bi-regional Meeting (Virtual) of National Influenza Centres (NICs) and influenza surveillance in WHO South-East Asia and Western Pacific Regions from 9 to 11 September 2025. Read morePUBLICATIONSImplications of influenza A virus surveillance in Southeast Asian Region countries: a scoping review of approaches for the surveillance of swine influenza viruses at human-swine interfaces. Read moreAnnual South-East Asia regional meeting on operational planning for implementation of global influenza strategy and preparedness for respiratory viruses in 2025 Report on annual review meeting of implementation of pandemic influenza preparedness (PIP) partnership Contribution (PC) Fund in South-East Asia Region and the simulation exercise on PRET Status of influenza in who south-East Asia Region
How CVS Health test-drives better care experiences using generative agents
CVS Health uses generative agent simulations to test customer experiences, uncover behavioral insights, and design better care journeys before real patients feel the impact.
Simulations shed light on how snowman-shaped body in Kuiper belt may have formed
Research adds weight to theory Arrokoth’s two lobes produced by gravitational collapse – and reveals process
Accelerating Science with Digital Twins
Researchers at Berkeley Lab are building digital twins that link physical systems with AI-informed models to accelerate discovery.
2026 Genesis Invitational odds, predictions, favorites, field: PGA picks this week from 10,000 simulations
Mike McClure's model has simulated the Genesis Invitational 2026 10,000 times and revealed best bets for this Signature Event
Tracking global water circulation using atomic fingerprints
Scientists have developed a powerful new way to trace the journey of water across the planet by reading tiny atomic clues hidden inside it. Slightly heavier versions of hydrogen and oxygen, called isotopes, shift in predictable ways as water evaporates and moves through the atmosphere. By combining eight advanced climate models into a single ensemble, researchers created the most accurate large-scale simulation yet of how water circulates worldwide.
Accelerating science with AI and simulations
For more than a decade, MIT Associate Professor Rafael Gómez-Bombarelli has used artificial intelligence to create new materials. Now, he believes AI is poised to transform science in ways never before possible.
NFL Mock Draft 2026: Eagles 7-round simulation, Version 4.0
How would you grade this projection?
Scientists discover hidden deep-Earth structures shaping the magnetic field
Deep inside Earth, two massive hot rock structures have been quietly shaping the planet’s magnetic field for millions of years. Using ancient magnetic records and advanced simulations, scientists discovered that these formations influence the movement of liquid iron in Earth’s core. Some parts of the magnetic field remained stable over vast stretches of time, while others changed dramatically.
It May Be Safe to Nuke an Earthbound Asteroid After All, Simulation Suggests
Hollywood was right all along.
NFL Mock Draft 2026: Eagles 7-round simulation, Version 3.0
How would you grade this projection?
From stars to simulations, counting down our 10 favorite science stories
We’re celebrating Texas Standard’s 10th birthday with a year of Top 10 lists.
Testing the system: regional simulation exercises advance global health security
These exercises – South-East Asia Regional Practice of All-Hazard IHR Event Communications (SAPHIRE), Joint Assessment and Detection of Events (JADE) in the European Region, and IHR Exercise Crystal in the Western Pacific Region – bring together National Focal points (NFPs), who play a critical role under the IHR in safeguarding global health security. These annual exercises stress-test IHR event communication procedures and contingency plans, demonstrating a sustained commitment by countries to strengthening preparedness and learning.
NYC Health + Hospitals Announces New Simulation Training for Emergency Department Providers Encountering Patients with Opioid Use Disorder - NYC Health + Hospitals
New York, NY – NYC Health + Hospitals today announced its new simulation training for Emergency Department providers to strengthen how they engage with
Meet the founder of this company that uses AI-powered gamification for incident response
IR Game mirrors tabletop training but compresses it into a high-intensity, guided simulation.
UNCG Nursing Innovates Training to Improve Maternal Health Outcomes
Professors Tammy Hall and Lisa Anders use simulation training to address maternal mortality disparities at UNCG.
The magnetic secret inside steel finally explained
For years, scientists noticed that magnetic fields could improve steel, but no one knew exactly why. New simulations reveal that magnetism changes how iron atoms behave, making it harder for carbon atoms to slip through the metal. This slows diffusion at the atomic level and alters steel’s internal structure. The insight could lead to more efficient, lower-energy ways to make stronger steel.
NASA’s SunRISE SmallSats Ace Tests, Moving Closer to Launch
The small satellites passed a crucible of tests, ending with a simulation of the intense vibrations of launch.
Scientists Preparing to Simulate Human Brain on Supercomputer
Researchers at the Jülich Research Centre in Germany are working on a simulation at the scale of the entire human brain.
How everyday foam reveals the secret logic of artificial intelligence
Foams were once thought to behave like glass, with bubbles frozen in place at the microscopic level. But new simulations reveal that foam bubbles are always shifting, even while the foam keeps its overall shape. Remarkably, this restless motion follows the same math used to train artificial intelligence. The finding hints that learning-like behavior may be a fundamental principle shared by materials, machines, and living cells.
Teaching Legislative Politics Through a Game: Active Learning, Assignment Co-creation, and Assessment Across Modalities -
Teaching Legislative Politics Through a Game: Active Learning, Assignment Co-creation, and Assessment Across Modalities By Jennifer L. Clemens, University of Wisconsin–Parkside and Michael A. Hansen, University of Turku This article examines a legislative politics simulation [...]
This simple math trick could transform earthquake science
Earthquakes happen daily, sometimes with devastating consequences, yet predicting them remains out of reach. What scientists can do is map the hidden layers beneath the surface that control how strongly the ground shakes. A new approach speeds up complex seismic simulations by a factor of about 1,000, making risk assessments far more practical. While it won’t forecast the next quake, it could help cities better prepare for one.
A missing flash of light revealed a molecular secret
Scientists have found a way to see ultrafast molecular interactions inside liquids using an extreme laser technique once thought impossible for fluids. When they mixed nearly identical chemicals, one combination behaved strangely—producing less light and erasing a single harmonic signal altogether. Simulations revealed that a subtle molecular “handshake” was interfering with electron motion. The discovery shows that liquids can briefly organize in ways that dramatically change how electrons behave.
Scientists replayed evolution and found a surprise
Environmental change doesn’t affect evolution in a single, predictable way. In large-scale computer simulations, scientists discovered that some fluctuating conditions help populations evolve higher fitness, while others slow or even derail progress. Two populations facing different kinds of change can end up on completely different evolutionary paths. The findings challenge the idea that one population’s response can represent a whole species.
Rams vs. Falcons prediction, odds: 'Monday Night Football' picks from 10,000 simulations
SportsLine's model simulated Los Angeles vs. Atlanta on 'MNF' 10,000 times and reveals its NFL picks today
100 Tennessee vs. Illinois simulations predict a close Music City Bowl
We're back to see EA Sports College Football 26's game prediction for Volunteers football vs. the Fighting Illini at Nissan Stadium.
This simulation reveals what really happens near black holes
Black holes are among the most extreme objects in the universe, and now scientists can model them more accurately than ever before. By combining Einstein’s gravity with realistic behavior of light and matter, researchers have built simulations that closely match real astronomical observations. These models reveal how matter forms chaotic, glowing disks and launches powerful outflows as it falls into black holes. It’s a major step toward decoding how these cosmic engines actually work.
Accelerating AI-Powered Chemistry and Materials Science Simulations with NVIDIA ALCHEMI Toolkit-Ops
Machine learning interatomic potentials (MLIPs) are transforming the landscape of computational chemistry and materials science. MLIPs enable atomistic simulations that combine the fidelity of…
"MicroProse and Geoff Crammond Reunite to Revive a Racing Legend"
MicroProse is excited to announce its renewed partnership with legendary game developer Geoff Crammond, creator of some of the most influential racing simulation games of the 1990s and early 2000s. Together, we are bringing back his groundbreaking racing game engine in a re-release of the original game technology, enhanced with updates, new tools, and Steam Workshop support, through the upcoming Geoff Crammond Racing series: GCR1, GCR2, GCR3, and GCR4, scheduled to release in 2026. These titles are being reintroduced under new names to reflect our independent direction and to ensure full separation from any official motorsport licensing. While the names are new, the core simulation engine, physics systems, and pioneering gameplay features developed by Geoff Crammond remain at the heart of each release. Geoff Crammond’s original work helped define the foundations of modern racing games, introducing innovations such as advanced real-time physics, highly configurable driving assists (including the ideal racing line, steering help and auto braking), realistic AI driver behaviour, deep telemetry analysis, comprehensive car setup systems, extensive wet weather system, and the list goes on. The GCR series proudly carries this legacy forward. Steam Workshop support will enable the community to create and share custom tracks, cars, liveries, teams, drivers and more. This reunion marks the beginning of a long-term partnership between MicroProse and Geoff Crammond, with further announcements in the series to come. MicroProse is officially back on the track at full speed. GCR1 Steam Page GCR2 Steam Page GCR3 Steam Page GCR4 Steam Page -- MicroProse, founded in 1982, is a world-famous simulation software company known for developing groundbreaking classics and new blockbuster titles such as Falcon, HighFleet, B-17 The Mighty Eighth, Ground Branch, Gunship and many more! Today, MicroProse is not only remaking select classics from its past but also paving the road to next-generation, state of the art games that will create the memories and experiences of new generations of gamers.
Chargers vs. Eagles prediction, odds: 'Monday Night Football' picks today from 10,000 simulations
SportsLine's model simulated Philadelphia vs. Los Angeles 10,000 times and revealed its 'MNF' best bets
Scientists Unveil the Most Realistic Black Hole Accretion Model Ever Created
Using cutting-edge algorithms and exascale supercomputers, researchers have created the most realistic simulations yet of matter flowing into black holes. Building on decades of research, a group of computational astrophysicists has reached an important breakthrough: they have created the most de
Scientists Finally Explain Mysterious “Impossible” Merger of Two Massive Black Holes
A detailed suite of simulations conducted by astrophysicists at the Flatiron Institute and their collaborators showed that magnetic fields can produce black holes with masses that were previously believed to be mostly unattainable. In 2023, astronomers recorded a dramatic cosmic event. Two unusua
How Data is Changing Science – Part 3: The Infrastructure Layer
In Part 1 of this series, we examined how science is beginning to change. Less fieldwork, fewer lab hours. More screens. More simulations. Instead of waiting for nature to do […]
CNN Academy Brings Real-Time Sports Newsroom Simulation to Georgia Tech
ATLANTA—CNN and the Georgia Institute of Technology today announced the launch of the CNN Academy Sports Journalism Bootcamp and Simulation, a two-week immersive program designed to prepare the next generation of storytellers for the fast-paced world of sports media.
Scientists say they may have finally solved the question of whether our universe is a simulation
New study claims to settle the mystery of whether reality is simulated.