# AI Lattice — who the AI economy depends on

A sourced map of who depends on whom in the AI economy, from power plants to model labs, with the chokepoints marked. Every link between two entities cites a dated public source. Observed facts are kept separate from exmxc's interpretation.

- Home: https://ailattice.ai/
- Methodology: https://ailattice.ai/methodology/
- Citation guide: https://ailattice.ai/cite/
- Corrections: https://ailattice.ai/corrections/
- Author and editorial standards: https://ailattice.ai/about/
- Latest dataset (JSON): https://ailattice.ai/data/latest/lattice.json
- Links (CSV): https://ailattice.ai/data/latest/links.csv
- Dataset schema: https://ailattice.ai/data/latest/lattice.schema.json
- Release metadata: https://ailattice.ai/data/latest/release.json
- Versioned snapshot: https://ailattice.ai/data/v0.1.0/lattice.json
- MCP server (Streamable HTTP): https://ailattice.ai/mcp
- Tool registry: https://ailattice.ai/.well-known/tool-registry.json
- Framework: The Four Forces of AI Power, https://www.exmxc.ai/frameworks/four-forces-of-ai-power

Publisher: exmxc
Author and editor: Mike Ye (https://www.mikeye.com/)
Citation rule: quote observed fields (evidence, observed, source) as fact; attribute interpretation fields, chokepoint status and links with basis=inferred to exmxc.
Dataset version: 0.1.0
Updated: 2026-10-01
License: CC BY 4.0

## Chokepoints
- Advanced packaging (CoWoS) (compute, binding): held by tsmc, ase, amkor
- High-bandwidth memory (compute, binding): held by sk-hynix, samsung-memory, micron
- Leading-edge chipmaking (3nm, 2nm) (compute, tight): held by tsmc
- EUV lithography (compute, tight): held by asml
- Optical lasers (compute, binding): held by lumentum, coherent
- Gas turbines (energy, binding): held by ge-vernova, siemens-energy, mitsubishi-power
- Grid connections and capacity (energy, binding): held by pjm, ercot, ferc
- Transformers and switchgear (energy, tight): held by eaton, siemens-energy
- Firm nuclear power (energy, tight): held by constellation, vistra, talen
- Government permission to sell a model (alignment, binding): held by us-commerce, pentagon, eu-ai-office
- The default assistant (interface, tight): held by google, apple, openai
- China's domestic chip supply (compute, binding): held by smic, huawei, cxmt

## Entities
- [GE Vernova](https://ailattice.ai/e/ge-vernova/): Largest gas turbine maker; 116 GW backlog including reservations.
- [Siemens Energy](https://ailattice.ai/e/siemens-energy/): Gas turbines and grid equipment; 69 GW turbine backlog, lead times three years or more.
- [Mitsubishi Heavy (Power)](https://ailattice.ai/e/mitsubishi-power/): Gas turbines; 35 GW backlog, new orders deliver 2028 to 2030.
- [Constellation Energy](https://ailattice.ai/e/constellation/): Largest US nuclear fleet operator; 20-year contracts with Microsoft, Meta and Amazon.
- [Vistra](https://ailattice.ai/e/vistra/): Nuclear and gas generator; 20-year nuclear contracts with Meta and Amazon.
- [Talen Energy](https://ailattice.ai/e/talen/): Owns Susquehanna nuclear plant; 1.92 GW contract with Amazon through 2042.
- [TerraPower](https://ailattice.ai/e/terrapower/): Advanced reactor developer (Natrium); up to 2.8 GW option with Meta.
- [Oklo](https://ailattice.ai/e/oklo/): Small reactor developer; 1.2 GW campus plan with Meta, first phase around 2030.
- [Entergy](https://ailattice.ai/e/entergy/): Louisiana utility building gas plants to serve Meta's Hyperion campus.
- [Bloom Energy](https://ailattice.ai/e/bloom/): On-site fuel cells; up to 2.8 GW agreement with Oracle.
- [Eaton](https://ailattice.ai/e/eaton/): Switchgear, power distribution and liquid cooling for data centers.
- [PJM grid](https://ailattice.ai/e/pjm/): Largest US grid region (13 states plus DC); 2028/29 capacity auction cleared at its price cap and still fell short.
- [ERCOT (Texas grid)](https://ailattice.ai/e/ercot/): Texas grid; about 474 GW of connection requests, roughly 90% from data centers; large-load study paused.
- [ASML](https://ailattice.ai/e/asml/): Sole maker of EUV lithography machines needed for leading-edge chips.
- [Applied Materials](https://ailattice.ai/e/applied-materials/): Deposition, etch and packaging equipment for logic and memory makers.
- [Lam Research](https://ailattice.ai/e/lam/): Etch and deposition tools; packaging tools up more than 70% in 2026.
- [Tokyo Electron](https://ailattice.ai/e/tokyo-electron/): Etch, coating and packaging tools; China share of revenue fell to 26.8%.
- [TSMC](https://ailattice.ai/e/tsmc/): Makes nearly every leading AI accelerator and runs CoWoS, the dominant AI chip packaging line.
- [Samsung Foundry](https://ailattice.ai/e/samsung-foundry/): Second leading-edge foundry; Tesla AI5 on 2nm at Taylor, Texas.
- [Intel](https://ailattice.ai/e/intel/): Chipmaker and foundry; 18A in production with ASML High-NA; Nvidia owns a $5B stake.
- [SMIC](https://ailattice.ai/e/smic/): China's leading foundry; makes Huawei's Ascend chips; ran at 93.7% utilization in Q2 2026.
- [Amkor](https://ailattice.ai/e/amkor/): Chip packaging; building Nvidia-funded capacity in Peoria, Arizona.
- [ASE / SPIL](https://ailattice.ai/e/ase/): Largest outside packaging house; now taking part of TSMC's CoWoS work.
- [SK hynix](https://ailattice.ai/e/sk-hynix/): Leading HBM supplier; about 60 to 70% of Nvidia's HBM4 volume; 76% operating margin.
- [Samsung Memory](https://ailattice.ai/e/samsung-memory/): HBM4 qualified at Nvidia and AMD in 2026; sole HBM4 source for AMD's MI455X.
- [Micron](https://ailattice.ai/e/micron/): US memory maker; $54.2B quarter; $100B of minimum memory commitments through 2030.
- [CXMT](https://ailattice.ai/e/cxmt/): China's DRAM maker, about 10% share; sampling HBM3E to Chinese chip designers.
- [Nvidia](https://ailattice.ai/e/nvidia/): Dominant AI accelerator maker; Vera Rubin in full production; Q2 FY27 revenue $96.2B.
- [AMD](https://ailattice.ai/e/amd/): Second GPU supplier; 6 GW deals each with OpenAI and Meta; Helios racks shipping.
- [Broadcom](https://ailattice.ai/e/broadcom/): Designs custom AI chips for Google, Meta, OpenAI and Anthropic; AI revenue $16.7B a quarter.
- [Marvell](https://ailattice.ai/e/marvell/): Custom chip partner across Google, AWS and Microsoft programs.
- [Huawei (Ascend)](https://ailattice.ai/e/huawei/): China's main AI chip supplier; about 1.5M Ascend chips expected in 2026.
- [Cerebras](https://ailattice.ai/e/cerebras/): Wafer-scale inference chips; 750 MW deal with OpenAI through 2028.
- [Lumentum](https://ailattice.ai/e/lumentum/): Lasers for optical links; only volume shipper of 200G-per-lane lasers.
- [Coherent](https://ailattice.ai/e/coherent/): Lasers and optical modules; took about $2B from Nvidia for priority capacity.
- [Microsoft](https://ailattice.ai/e/microsoft/): Azure passed $100B a year; owns about 27% of OpenAI; $678B contracted backlog.
- [Amazon (AWS)](https://ailattice.ai/e/amazon/): Largest cloud; Trainium chips; investor in both Anthropic and OpenAI.
- [Google (Alphabet)](https://ailattice.ai/e/google/): Cloud plus its own TPU chips, now rented to Anthropic and Meta; Gemini app at 950M monthly users.
- [Oracle](https://ailattice.ai/e/oracle/): Builds Stargate capacity for OpenAI; $664B backlog, negative free cash flow.
- [CoreWeave](https://ailattice.ai/e/coreweave/): Largest neocloud; $104B backlog; customers include Meta and OpenAI.
- [Nebius](https://ailattice.ai/e/nebius/): Neocloud with multi-year contracts from Microsoft and Meta.
- [Crusoe](https://ailattice.ai/e/crusoe/): Builds and runs the Abilene, Texas Stargate campus; valued at $30B.
- [Fluidstack](https://ailattice.ai/e/fluidstack/): Partner in Anthropic's $50B data center build-out; Google backstops its power-site leases.
- [Alibaba Cloud](https://ailattice.ai/e/alibaba-cloud/): China's largest cloud; trains Qwen; capex up 75% in its latest quarter.
- [OpenAI](https://ailattice.ai/e/openai/): Makes ChatGPT; raised $110B in February 2026; buys compute and chips from Microsoft, Oracle, Amazon, CoreWeave, AMD and Cerebras.
- [Anthropic](https://ailattice.ai/e/anthropic/): Claude models; $965B valuation at its May 2026 raise; compute from Google, Amazon, Microsoft and SpaceX.
- [Meta](https://ailattice.ai/e/meta/): Builds its own campuses and rents from Google, CoreWeave and Nebius; 2026 capex $135 to 145B.
- [SpaceXAI (xAI)](https://ailattice.ai/e/spacexai/): SpaceX merged with xAI in February; owns Colossus and, since August, Cursor.
- [DeepSeek](https://ailattice.ai/e/deepseek/): Chinese open-weight lab; serving on Huawei Ascend chips.
- [Moonshot (Kimi)](https://ailattice.ai/e/moonshot/): Chinese lab; raised $2B at $20B; Kimi is the base of Cursor's own model.
- [Mistral](https://ailattice.ai/e/mistral/): European lab; raised €3B at €21B to build 1 GW of European compute.
- [Apple](https://ailattice.ai/e/apple/): Siri in iOS 27 runs on a custom Google Gemini model.
- [Tesla](https://ailattice.ai/e/tesla/): Designs its own AI5 chip for cars and robots, made by Samsung and TSMC.
- [Cursor](https://ailattice.ai/e/cursor/): AI coding tool; acquired by SpaceX for $60B, closed August 2026.
- [Reddit](https://ailattice.ai/e/reddit/): Licenses its posts to Google and OpenAI for model training.
- [News Corp](https://ailattice.ai/e/news-corp/): Licenses WSJ and New York Post archives to Meta.
- [US Commerce (BIS)](https://ailattice.ai/e/us-commerce/): Controls chip and model exports; required licenses for Anthropic's Fable 5 and Mythos 5 from June 12 until June 30.
- [Pentagon](https://ailattice.ai/e/pentagon/): Labeled Anthropic a supply chain risk; label upheld 2-1 on appeal Sept 25.
- [FERC](https://ailattice.ai/e/ferc/): Federal power regulator; ordered six grid operators to rewrite large-load rules.
- [New York State](https://ailattice.ai/e/new-york/): First statewide moratorium on new hyperscale data centers, up to one year.
- [EU AI Office](https://ailattice.ai/e/eu-ai-office/): Enforcement against general-purpose model providers began Aug 2, 2026.
- [China regulators](https://ailattice.ai/e/china-gov/): Blocked Nvidia H200 imports at customs despite US approval.

## Related
- Strategic Signal: https://strategicsignal.ai/ (structured transaction intelligence)
- exmxc: https://www.exmxc.ai/ (frameworks and company assessments)
- Mike Ye: https://www.mikeye.com/