Jesse Hawkins · AI and chips, measured
Catch up on the week in AI and chips The seven days to 9 Nov hold 272 events on the Tape, 3 scheduled releases and no new highs and lows.
Forget my visits Follow AI and chip filings, rules and models as they land The Tape → 3 Tue 4 Wed 5 Thu 6 Fri 7 Sat 8 Sun 9 Mon
Filings20 Texas Instruments filed an 8-K: other events KLA filed an 8-K: shareholder vote, other events Super Micro Computer filed a 10-Q Meta filed an 8-K: other events Micron filed an 8-K: officers or directors Lam Research filed an 8-K: shareholder vote, other events Alphabet filed an 8-K: other events Microsoft filed an 8-K: officers or directors Oracle filed an 8-K: other events Arm Holdings filed a 6-K Qualcomm filed a 10-K Hewlett Packard Enterprise filed an 8-K: Regulation FD Qualcomm filed an 8-K: results Arm Holdings filed a 6-K Microsoft filed an 8-K: other events GlobalFoundries filed a 6-K GlobalFoundries filed a 6-K AMD filed a 10-Q Super Micro Computer filed an 8-K: results, other events AMD filed an 8-K: results, Regulation FD Rules and orders3 Extension of the Suspension of Expansion of End-User Controls for Affiliates of Certain Listed Entities Agency Information Collection Activities; Submission to the Office of Management and Budget (OMB) for Review and Approval; Comment Request; Section 232 National Security Adjustments to Imports of Specialty Pharmaceuticals and Associated Ingredients Agency Information Collection Activities; Submission to the Office of Management and Budget (OMB) for Review and Approval; Comment Request; Application Process To Waive the Import Qualification Criteria Measures To Restrict Stockpiling of Polysilicon and Polysilicon Derivatives Models1 nvidia/MetricTrail Papers52 BitDecoding-FP4: Native FP4 KV Cache Decoding via Algorithm-System Co-Design Atik: A Shared-Mesh Transformer Accelerator for RISC-V Memory as a Temporal Resource: Lifetime Contracts for Multi-Retention Systems Robust & Scalable Website Fingerprinting via Minimal DRAM Traffic TwinHash: Two-Sided Object Identity Management for Time- and Energy-Efficient Agentic Workflows VFlash: Fast, Accurate Visual Planning via Flash Primitive Decoding SLA: Scaling Spatial-Language-Action Model via Continual Pre-training From Autonomous Agents to Cognitive Symbionts: Toward AI-Augmented Human Intelligence Gain-Dependent Entropy Characterization of a Zener Avalanche Noise TRNG via Dual NIST Standards: Toward Low-Cost Hardware Entropy for ICS Deployments CPU Micro-Architectural Characteristic Sensitivity Study on CPU-Side Agentic Task Latency Enabling Retraining-Free Accurate Inference in Analog Compute in Memory Accelerators Scaling Up Analog CIM: A Simulation Framework for Topology-Aware Mapping SensorGuard-RV: Sensor-to-Register Data Protection for Resource-Constrained RISC-V Devices Should You Split Decode? Disaggregation Across Future Models and Accelerators Zirah: A Post-Quantum True Random Number Generator Hardware Signer With an Avalanche Noise MCUFlowNet: Hardware-Aware Neural Architecture Search for Optical Flow on Microcontrollers CAMP: Context-Aware Mapping and Placement for LLMs on 3D-IMC Arrays OpenRDMA: A GPU-Native Multi-Path Reliable RDMA Transport Architecture Memory Fit: Architectural Support for Page Placement by Learning Models of Good Behavior SAFTAC: Simulation-Augmented Fine-Tuning of Open-Source LLMs for Analog Circuit Design A Synchronized Tri-Modality Wireless System for Multi-Point Non-Contact Cardiac Monitoring Rule-Guided and Trajectory-Aware Repair for LLM-Based HLS Generation Energy Measurement and Optimization of TinyML in Resource-Constrained Embedded Systems: A Systematic Mapping Study Turns Matter: Advancing Streaming Video RL via Turn-Level Credit Enabling AWS Cloud FPGA Development Using High-Level Vitis HLS Energy Efficient Near-Threshold Voltage Massive MIMO Computing Data Center Agglomeration and Power Infrastructure Capacity in the United States FPTR: Feasibility-Preserving Transactional Refinement for Deadline-Constrained Joint Beam and Resource Allocation Hardwired Distributed Processing Units for Inference: Neural Decisions at the Sensor that Free the Host TauLang: A Language and Compiler for Space-Time Programming on Statically Scheduled Accelerators PIMROP: A Feasibility Study of Return-Oriented Programming on Commercial Processing-in-Memory Architectures Phantom: Temporal Caching for Diffusion Models CHARM+: Cross-Hardware Attention with Re-Merge Multistream Mechanism RoadBlock: Rethinking GPU Tensor Core Microarchitecture for Emerging Microscaling Format Support Are Frontier VLM Agents Ready to Be Robot Generalists? An Empirical Study with the Embodied Agent Arena Route Anything: A Study of Model and Policy Routing for General-Purpose AI Systems Scheduling Data to Compute: Hierarchical Data Pipelining for LLM Training on AMD GPUs DynaMite: Dynamic Measurement Scheduling To Accelerate Fault-Tolerant Quantum Programs Workload-Static Optical Circuit Switching For Scalable LLM Inference Decode GSPIM: A Temporal-Locality-Aware 4DGS Accelerator for PIM-Enabled Edge Devices A RISC-V Simulator for the Tenstorrent Tensix Architecture Deep Learning Compilers: Architecture, Optimization, and Empirical Characterization UMLoad: Fast On-Demand Model Loading on Unified Memory Architecture GC-MoE: Overlapping Expert-Parallel Token Dispatch with FP8/FP4 Grouped GEMM via CUDA Green Contexts HyPAIC: Personal AI Computer with High Bandwidth Flash for High Performance Trillion-Parameter LLM Inference CoworkOS:Scaling Agent Coworkers for Real-world Automation Wave Forcing: Towards Speed-of-Light Streaming Video Generation Advances in Agentic Computer Graphics CUA-Master: Scaling Computer-use Agents through Open-World Recursive Improvements FrameForge: Harnessing Priors of Video Foundation Models for Unified Image Generation R^2Orch: Recursive Self-Improvement for Agent Orchestration via Self-Play within a Skill-Internalization Loop REA: A Routed Embodied Agent Runtime for Asynchronous Brain-Cerebellum Control Labs and press156 CTO Interview with Ed Plowman of Imagination Technologies Can Jensen Keep His Job? Fable 5.2 and Jagged AGI ChatGPT Enterprise now available in AWS Marketplace Golden Testcases You Can Check With a Calculator What Frontier Labs Buy: A Data Deep Dive with Handshake AI — Garrett Lord & Team Curvilinear OPC and next-gen photomask optimization Nemotron-Personas-Finance-Japan: A Synthetic Dataset Capturing Japan's Demographics and Financial Context Advancing Integrated Optics: Celestica & Broadcom Interview at OCP Global Summit Who Owns the Voice in Your Head? Powering AI Data Centers: Rethinking Energy Infrastructure for the Next Wave of AI Growth Practice Makes Precise: At CoRL, NVIDIA and Partners Reveal How Robots Rehearse Delicate Work in Simulation Still Looking for Diamonds Designing Smarter Security and Portable Camera Systems Control Room: A central system to track and stop model workloads across our research infrastructure Your Experts Have Partners. There's HOPE for Keeping Them Together NVIDIA Announces New Planned Investments in Korean Startups at the MSIT-NVIDIA AI Startup Partnership Forum Kuken Industries searches 70 years of drawings in minutes CodeRabbit scales agentic software development and ships 20x more DeepX Tapes Out Second-Gen Chip, Projects 6.7× Performance Gain With LPDDR5x-PIM Korea Advances Physical AI Across Industries With NVIDIA Korea's AI Ecosystem Advances and Secures Agentic AI With NVIDIA From Chips to Skills, NVIDIA and Korea Expand Sovereign AI Collaboration NVIDIA Dynamo Recipes Bring Korea's Leading Open Models to Production NVIDIA, Rebellions and SK Telecom Announce Collaboration to Advance Sovereign AI Infrastructure With NVIDIA NVLink Fusion The Web Is More Than Enough, by Paul Ford Hide your slop PR descriptions behind parse_number() as a SQLite extension: C vs Rust How Stanford Teaches AI-Native Engineering — Mihail Eric, Monaco [One Million AI Engineer Special] How Stanford Teaches AI-Native Engineering — Mihail Eric, Monaco A Message to Our Company CEO Interview with Mark Rushworth of Finchetto Statement from NVIDIA Board of Directors CEO Interview With Trey Closson of Amplio CEO Interview with Albert Liu of Kneron CEO Interview with Moshiel Biton of Addionics CEO Interview with Brett Walker of Electroninks CEO Interview with Tallat M. Shafaat of Vectara CEO Interview with Jun Yuan of Arcas-Tech CEO Interview with Pratap Ranade of Arena Physica datasette #3046: Don't serve link: rel=next HTTP headers to Firefox for HTML pages Terry Pratchett's Study We Visited China's Stargate in Inner Mongolia [AINews] Anthropic IPO: Whither Now? The Countdown to Q-Day Has Begun Cosmic at SEMICON Europa 2026: Advancing High-Complexity Semiconductor Test Solutions GPT-6 Astra Covertly Underperforms on Safety Research Tasks ChipNexus Brings Agentic AI into Established Analog Design Flows Quoting Nilay Patel Delivering Silicon Systems Through AI Disruption, Part 2: The Conductor How NVIDIA Open Source Tools Are Helping Researchers Build Next-Generation 6G Networks Controlling Biology - Fabian Theis, Helmholtz Munich IBM, AMD Mix Quantum and AI at U.S. National Lab ToyBlocks: A controlled world for robot learning and evaluation [AINews] Inkling 2 Small: 17A-594B trained in 68 hours on 12T tokens, beating Muse Spark Europe's Chip Industry: Perfect Storm or Perfect Timing? Partnering with Hacktron to find vulnerabilities in OpenAI's systems Persistently High DRAM Prices Drive Adoption of Reballed Memory, with Smartphone Penetration Expected to Reach 7–9% in 2027, Says TrendForce Project EVE: Claude plays MMORPGs Measuring how AI agents handle everyday wet-lab science The components of agent memory Teaching AI to build scientific models Egyptian and Nigerian English join the Open ASR Leaderboard EE Times Magazine – November 2026 The benchmark where AIs have to lie to win Vera Delivers the Best of Both Worlds: Highest Throughput and Fastest Agents Interview with David Tester VP of Engineering Mach42 Cult.fit builds operational muscle with Codex New NVIDIA Cosmos 3 Open Models for Interactive World Simulation Infineon Leaders Andreas Urschitz and Maher Matta on a Decade of OktoberTech Infineon's Johannes Schoiswohl on the GaN Market at OktoberTech The Iron(wood) Throne: Google's TPU Stakes Its Claim on Inference Infineon's Steve Tateosian Talks IoT, Consumer, and Industrial MCUs at OktoberTech AI #193: Progress May Dramatically Accelerate From PyTorch to Product: Accelerating Edge AI Through Open Ecosystems Glass Is Becoming an AI Infrastructure Platform Accelerate Analog Design with Flexible, AI-Native Workflows Quoting Paul Ford Gemini Omni 1.2 Flash: Fast and cost-effective with longer clips Australian universities are creating AI-native graduates Gemini Omni 1.2 Flash: Fast and cost-effective with longer clips Nvidia, Municipal Leaders Frame AI as Critical Urban Infrastructure Ceva and Mobilint Bring Physical AI Closer to Everyday Devices The Wizard of Open Weights: Chris Alexiuk Makes AI Easy With Open Models More Performance From Every Byte: How Full-Stack Codesign Solves AI's Memory Challenge 'Hogwarts Legacy' Gets a Magical NVIDIA DLSS 5 Upgrade on GeForce NOW The $725 Million Chip Development: Why Execution Risk, Not IP Cost, Now Dictates Advanced-Node Survival GPT‑6 Cyber: Advancing the frontier of aligned cybersecurity agents Why Magnetic Sensor Selection Has Become a System Architecture Decision Merge builds product and business tools with OpenAI technology What Makes a Quantum Company Valuable? [AINews] It's Swapping Season Setting a new standard for responsible AI infrastructure Flipsnack grows product signups with conversion optimization A Closer Look at Phonon TTS. Launching High-Quality On-Device Speech That Tops the Benchmarks. Franklin Templeton uses OpenAI to support investment research Build Enterprise Agents, Workflows & Apps in Mistral Studio How scientists use Claude Chiplets: Meta-Scale Chiplet Velocity Accelerating AI SoC Development to Match the Rapid Evolution of AI Model CTO Interview with Taylor Barton of Certus Semiconductor Silicon Photonics Hits the Big Time [AINews] Behind Latent Space: Kill My SaaS, Jev & AI in 2026 | AIEi Shanghai Generating Chip Subsystems Automatically: A New Answer to Manage Chip Design Complexity Dharma Field Extractor: Any Document In, Structured Fields in JSON Out ABLIC's Small, Smart, and Simple Strategy for High-Value Analog Semicondutors How big is the open model economy today? Reimagining 3D IC Design: From Architecture to Signoff datasette-llm 0.1a10 GSA Honors Sanjay Mehrotra as Memory Takes Center Stage in Computing STMicroelectronics Expands STM32 Roadmap for Cloud AI and Physical AI OpenAI to acquire Pierre Computer Company, Inc. NIST Sets a Clearer Course for the Post-Quantum Security Transition Quantum Roadmaps Are Finally Meeting Reality A thousand fewer paths to root Daybreak Red is now available in AWS GovCloud (US-East) datasette-apps 0.3a0 Introducing Images 2.6 Investigating vault particles with Claude DevRev delivers autonomous self-service with GPT-6.1 Sol Is There Room for Small-Scale OSATs in India's Chip Packaging Boom? datasette-agent 0.5a0 llm 0.37.1 Elon Is Doing It Again: SpaceX Will Be The World's Biggest AI Cloud CTO Interview with Ed Plowman of Imagination Technologies U.S. Manufacturing Activity Gains Momentum in October IP update from Alphacore at DAC 2026 StateBench: Testing how language models track state changes over long documents Agentic cheminformatics with Google Antigravity OpenAI Updates Its Model Spec AI is destroying my CS degree Palette: teaching a small model to build decks in your design language Qwen 3.8 27B on my laptop can interpret model weights, too Ceva's next NeuPro generation opens new possibilities across AI AI Is Changing How Semiconductor Fabs Work — and Engineers Are Moving Faster Because of It Accepting Orders for the LED Light-Sourced Projection Aligners Bringing Design Data Management into the Developer Workflow AI Hardware's Hidden Cost: Between Orders and Output How Allegro MicroSystems Increased OEE by Several Percentage Points with yieldHUB Live Five Things You Didn't Know About Co-Packaged Optics Accelerating engineering for the physical world Marvell's Acquisition of Lumentum to Unite Optical and Electronic Capabilities, Potentially Challenging Broadcom's Market Leadership, Says TrendForce Marvell Snaps Photonics Chipmaker Lumentum to Bolster AI Connectivity Heimstaden is changing residential real estate management with AI How Cisco defends at machine speed with OpenAI Cities Worldwide Build and Scale AI Services With NVIDIA Physical and Agentic AI [AINews] LLMs can read & write neural network weights: "Weight Space Programming" cuts Autoresearch out of the RSI loop I discovered a sound that leaks your exact location (and it's everywhere) - YouTube Macronix Builds Hardware Security into NOR Flash Ahead of EU Cyber Resilience Act October sponsors-only newsletter OpenAI appoints Jill Hazelbaker as Chief Corporate Affairs Officer CWE-bench v2: More coverage, quality, and transparency Transformers Model Integration Contracts llm 0.37 Building AI for Lawyers - Jacob Lauritzen, CTO of Legora Building Legal Agents When You Can't See the Data - Jacob Lauritzen, CTO of Legora Progress on provable inference for large language models Discussion40 Claude Fable 5.2 and Mythos 5.2 Google Gemini routes personal Drive files and emails to human annotators ChatGPT Desktop forwards your SSH agent to remote machines I'm not convinced the AI bubble exists Nvidia CEO Jensen Huang's letter to staff about affair Nvidia Board of Directors Is Investigating Jensen Huang How I Fight AI Brain Fry Goldman Sachs is pulling Claude Code from its bankers and traders Llama-modchip: 284B DeepSeek on 64GB RAM, no GPU Goodbye Claude, hello Pi Claude Code's "Don't Ask Again" exposes passwords in 1,122 public GitHub files The end of Microsoft Research's AI Frontiers group AI is dramatically accelerating AI research Anthropic IPO buyers see valuation as low as $1.1T Gemini Omni 1.2 Flash I struggle to believe that AI R&D is ingenious Cuban: Facebook knowingly runs AI deepfake scam ads; docs say $16B fraud revenue Claude Code sends your email address to websites Hungary: Communist-era files released, up to 60k former agents revealed GPT-6 Astra covertly underperforms on safety research Ripple-65B Preview: U.S. Open-Weight SOTA on Agentic Coding Progress on provable inference for large language models OpenAI to acquire Pierre Computer Company Ask HN: What's your current AI agentic setup? US voters roundly reject data centers in ballot measures AI is forcing us to pay off 40 years of technical debt Geoffrey Hinton: it might turn out we were better off without AI Tell HN: Gemini trains on chat history, even for subscribers (personal accounts) Nvidia RTX Spark encrypted firmware slows inference speed to 35% on Linux The Little Book of Generative AI Foundations: An Intuitive Mathematical Primer US tech giants' hidden AI debt swells 90% to $3.2T in 6 months How I use AI (Oct 2026) Qwen 3.8 27B on my laptop can interpret model weights, too Postman silently auto-installs AI plugins and attempts to share all API data Google will turn off search for schools that reject AI OpenAI seems to be silently re-enabling model training on Codex chats The 'Astra can read and write weights to improve Qwen' post is misleading Nvidia – open-source auto-ontology Intel stock collapse worsens as $3.5T chip rout spreads Scaling Git at OpenAI Filings, rules, models, papers, lab posts and discussion: one tick an event, at the time its source gave. A taller, brighter tick drew more attention. Each event linked to its source · every source read every 6 hours, last at 9 Nov 2026, 22:42 UTC · Data: SEC EDGAR · Federal Register · eCFR · the White House · Hugging Face · arXiv · Hacker News · each lab’s own feed
Sign up free to use the Tools Every chart, map and record is open to read. A free account also opens the Tools, and sign-up takes under a minute with no password to remember.
Already a member? Sign in
Jesse Hawkins · The global AI buildout
Today Oct 2026
18.2 GW is what the world's largest AI data centers draw today
drawn by the world's 95 largest AI campuses, about 43% of the world's AI computing capacity
The power of 14.8 million US homes.
Online 74
Building 21
Data-centre power use, 2025 485 TWh
AI compute worldwide ~20.0M H100e
Data: Epoch AI , CC BY 4.0 · IEA, CC BY 4.0 · company filings · cloud providers ·
by country ·
sources
Open the Buildout →
A globe of the global AI buildout. Each light is one large AI campus, its area the campus's power; rings are AI supercomputers and small squares are public cloud regions.
Home power price , change since 2021, median state · brighter tiles have more AI capacity against demand · The comparison does not show cause.
Data centers 84% of tracked AI data-center power sits in the United States Compute 86% of tracked AI computing power sits in the United States Chip plants 52% of listed chip plants are in Japan, the US and China Sales +57% in a year for September 2026 sales of TSMC and 8 AI server makers, against +33% for the other 1,079 companies on Taiwan’s main board Imports 5.7× US imports of servers, chips and computer parts from Taiwan, Mexico and South Korea, the twelve months to August 2026 against four years earlier Data: Epoch AI, CC BY 4.0 · Wikipedia's contributors, CC BY-SA 4.0 · Taiwan FSC via TWSE, OGDL 1.0 · U.S. Census Bureau, public domain · sources on each page
2023 · from May 4,247 2024 · derived 12,742 2025 54,836 2026 · to Sep 120,136 Announced job cuts, AI given as the reason · Data: Challenger, Gray & Christmas · New York State Department of Labor · SEC EDGAR · no company named
2024 2025 2026
proprietary
open weights · Each block is one model’s time in first place. The board began in May 2023 for Text; the others later.
First place on four Arena leaderboards · Data: LMArena, CC BY 4.0