AI in China
17.5% of people of working age in China were using AI tools in 2026-Q2, by Microsoft's estimate, behind 66 of 147 economies. 7 instruments relevant to AI in Chinese law and policy are indexed, 7 of them binding. Epoch AI lists 190 notable models with a developer in China, and 5 AI datacentre sites are recorded there in Epoch AI's hub.
- 17.5% of the working-age population using AI tools, 2026-Q2 AI Diffusion Report: AI User Share Microsoft AI for Good Lab Derived from anonymised Microsoft telemetry, then adjusted for OS/device market share, internet penetration and a modelled mobile-to-desktop ratio. It therefore measures Microsoft-visible AI use, scaled, not observed total AI use. Systematically less reliable where Microsoft platform share is low, and it cannot see AI embedded in products users do not recognise as AI. Cite Microsoft AI for Good Lab, "AI Diffusion Report: AI User Share". https://github.com/microsoft/ai-diffusion-report. Retrieved 28 October 2026 via AI Safety Tracker, https://aisafetytracker.org/china/.
- 7 of 7 indexed instruments that are binding law AI Law and Policy Index This site Our own reading of primary instruments, with a link to each. Not legal advice, not a compliance checklist and not exhaustive. Instruments that never mention AI are included where they apply to it, which is most of the Australian list. Statuses and commencement dates go stale; every record carries its review date. Cite AI Safety Tracker, "AI Law and Policy Index". https://aisafetytracker.org/policy/. Retrieved 23 October 2026.
- 9 of 10 governance readiness points AI Governance Readiness Index This site Our own editorial coding from primary legal sources, not an objective measurement. Every score cites a specific instrument and article. "Readiness" means the machinery exists to govern AI. It is emphatically not a judgement that the resulting rules are good, proportionate, or well enforced. Cite AI Safety Tracker, "AI Governance Readiness Index". https://aisafetytracker.org/methodology/governance-readiness/. Retrieved 23 October 2026.
- 2.1 GW of IT power planned at 5 sites in Epoch AI's hub, 731 MW of it operating AI Data Centers Epoch AI The largest AI sites Epoch has found, mostly in the United States. Epoch estimated in September 2026 that they hold 44% of the world's deployed AI compute, with a 90% interval of 23% to 81%. Power, compute and cost are Epoch's estimates from satellite imagery, permits and company statements; four in five IT power estimates fall within a factor of 1.4 of the true value. The owner is the owner of the chips, not necessarily of the building. Capacity after the date of retrieval is Epoch's projection. Coordinates are read from the hub's map page, which publishes them; the downloads do not. Cite Epoch AI, "AI Data Centers". https://epoch.ai/data/ai-data-centers. Retrieved 28 October 2026 via AI Safety Tracker, https://aisafetytracker.org/china/.
Law and policy
7 instruments that apply to AI in China, grouped by how binding each is. 7 are binding law, and 2 of those name AI or deepfakes in their titles. AI Law and Policy Index This site Our own reading of primary instruments, with a link to each. Not legal advice, not a compliance checklist and not exhaustive. Instruments that never mention AI are included where they apply to it, which is most of the Australian list. Statuses and commencement dates go stale; every record carries its review date. Cite AI Safety Tracker, "AI Law and Policy Index". https://aisafetytracker.org/policy/. Retrieved 23 October 2026.
| Binding instrument | Kind | Year |
|---|---|---|
| Measures for Labelling AI-Generated Synthetic Content | Rule under an Act | 2025 |
| Interim Measures for the Management of Generative AI Services | Rule under an Act | 2023 |
| Provisions on Deep Synthesis Internet Information Services | Rule under an Act | 2022 |
| Data Security Law | Act | 2021 |
| Personal Information Protection Law | Act | 2021 |
| Provisions on Algorithmic Recommendation of Internet Information Services | Rule under an Act | 2021 |
| Cybersecurity Law | Act | 2016 |
Data: the AI Law and Policy Index, read from each primary instrument. Not legal advice.
Use of AI tools
Share of the working-age population using AI tools, by reporting period, and the economies ranked either side of China in 2026-Q2. Microsoft's estimate from its own telemetry, scaled to each population. AI Diffusion Report: AI User Share Microsoft AI for Good Lab Derived from anonymised Microsoft telemetry, then adjusted for OS/device market share, internet penetration and a modelled mobile-to-desktop ratio. It therefore measures Microsoft-visible AI use, scaled, not observed total AI use. Systematically less reliable where Microsoft platform share is low, and it cannot see AI embedded in products users do not recognise as AI. Cite Microsoft AI for Good Lab, "AI Diffusion Report: AI User Share". https://github.com/microsoft/ai-diffusion-report. Retrieved 28 October 2026 via AI Safety Tracker, https://aisafetytracker.org/china/.
- 15.4% 2025-H1
- 16.3% 2025-H2
- 16.4% 2026-Q1
- 17.5% 2026-Q2
| Economy | 2026-Q2 | Rank | Relative magnitude |
|---|---|---|---|
| Türkiye | 18.9% | 62 | |
| Azerbaijan | 18.7% | 63 | |
| India | 18.5% | 64 | |
| Romania | 18.0% | 65 | |
| Mongolia | 17.9% | 66 | |
| China | 17.5% | 67 | |
| Guatemala | 17.3% | 68 | |
| Peru | 17.3% | 68 | |
| Kazakhstan | 16.9% | 70 | |
| Gabon | 15.7% | 71 | |
| Egypt | 15.5% | 72 |
Ranked within 147 economies. Bars are scaled to the highest share, United Arab Emirates at 73.3%. The estimate measures AI use visible to Microsoft and is least reliable where its platform share is low.
Governance readiness
9 of 10 points on a five-dimension rubric recording how much machinery exists to govern AI, not how strict it is. 1 of 31 nationally coded jurisdictions score higher and 0 score the same. AI Governance Readiness Index This site Our own editorial coding from primary legal sources, not an objective measurement. Every score cites a specific instrument and article. "Readiness" means the machinery exists to govern AI. It is emphatically not a judgement that the resulting rules are good, proportionate, or well enforced. Cite AI Safety Tracker, "AI Governance Readiness Index". https://aisafetytracker.org/methodology/governance-readiness/. Retrieved 23 October 2026.
| Dimension | Score | What the score records |
|---|---|---|
| Binding law | 3 of 3 | A horizontal instrument in force and applicable. |
| Oversight body | 2 of 2 | A regulator holding statutory powers over AI. |
| Evaluation capacity | 1 of 2 | Announced or embryonic. |
| Transparency duty | 2 of 2 | An explicit AI disclosure, labelling or automated-decision duty in law. |
| Strategy | 1 of 1 | A published national AI strategy. |
Scores as horizontal because the binding rules, taken together, reach AI services generally, even though they are a stack of administrative measures rather than one omnibus act: algorithmic recommendation, deep synthesis, generative AI services, and from September 2025 mandatory labelling of AI-generated content. The Cyberspace Administration of China registers algorithms and enforces. Article 24 of PIPL gives automated-decision rights. Cited for the score: Interim Measures for Generative AI Services (2023), Measures for Labelling AI-Generated Content (2025), Personal Information Protection Law (2021).
Data: the AI Governance Readiness Index. Against AI use in every jurisdiction: exposure against governance readiness
AI datacentres and compute
5 AI datacentre sites in Epoch AI's hub in China, by the IT power planned at the last stage of each site's timeline, with the power operating now. AI Data Centers Epoch AI The largest AI sites Epoch has found, mostly in the United States. Epoch estimated in September 2026 that they hold 44% of the world's deployed AI compute, with a 90% interval of 23% to 81%. Power, compute and cost are Epoch's estimates from satellite imagery, permits and company statements; four in five IT power estimates fall within a factor of 1.4 of the true value. The owner is the owner of the chips, not necessarily of the building. Capacity after the date of retrieval is Epoch's projection. Coordinates are read from the hub's map page, which publishes them; the downloads do not. Cite Epoch AI, "AI Data Centers". https://epoch.ai/data/ai-data-centers. Retrieved 28 October 2026 via AI Safety Tracker, https://aisafetytracker.org/china/.
| Site | Planned IT power | Now | Relative magnitude |
|---|---|---|---|
| VNET Bayin Ulanqab | 700 MW | 221 MW | |
| Envision Galaxy Ulanqab | 470 MW | none | |
| Huawei Wuhu | 457 MW | 157 MW | |
| Huawei Horinger | 346 MW | 184 MW | |
| Alibaba Zhangbei | 169 MW | 169 MW |
- 731 MW IT power operating at those sites now AI Data Centers Epoch AI The largest AI sites Epoch has found, mostly in the United States. Epoch estimated in September 2026 that they hold 44% of the world's deployed AI compute, with a 90% interval of 23% to 81%. Power, compute and cost are Epoch's estimates from satellite imagery, permits and company statements; four in five IT power estimates fall within a factor of 1.4 of the true value. The owner is the owner of the chips, not necessarily of the building. Capacity after the date of retrieval is Epoch's projection. Coordinates are read from the hub's map page, which publishes them; the downloads do not. Cite Epoch AI, "AI Data Centers". https://epoch.ai/data/ai-data-centers. Retrieved 28 October 2026 via AI Safety Tracker, https://aisafetytracker.org/china/.
- 320,000 H100 equivalents of compute at those sites AI Data Centers Epoch AI The largest AI sites Epoch has found, mostly in the United States. Epoch estimated in September 2026 that they hold 44% of the world's deployed AI compute, with a 90% interval of 23% to 81%. Power, compute and cost are Epoch's estimates from satellite imagery, permits and company statements; four in five IT power estimates fall within a factor of 1.4 of the true value. The owner is the owner of the chips, not necessarily of the building. Capacity after the date of retrieval is Epoch's projection. Coordinates are read from the hub's map page, which publishes them; the downloads do not. Cite Epoch AI, "AI Data Centers". https://epoch.ai/data/ai-data-centers. Retrieved 28 October 2026 via AI Safety Tracker, https://aisafetytracker.org/china/.
- 188 GPU clusters in Epoch AI's cluster records, 297 MW of known power GPU Clusters Epoch AI Compiled from public reporting; each row carries a certainty flag ranging from confirmed to speculative. Chinese clusters are systematically less well documented than US ones, so country totals understate opacity as much as they measure capacity. Cite Epoch AI, "GPU Clusters". https://epoch.ai/data/gpu-clusters. Retrieved 28 October 2026 via AI Safety Tracker, https://aisafetytracker.org/china/.
Hub power and compute are Epoch's estimates, and planned power is its projection from permits and announcements. Epoch's cluster records are a separate compilation, not reconciled with the hub, and count power only where it is known.
Data: Epoch AI
Notable models
190 models in Epoch AI's notable models list with an organisation in China among their developers, 60 of them from academic groups and 65 published since January 2025. The largest training compute recorded for any of them is 2.5e25 FLOP, for Doubao-pro. Notable AI Models Epoch AI Inclusion is editorial ("notability criteria"), not exhaustive. Compute figures are frequently estimates; check the per-row confidence and notes columns before treating any value as measured. Cite Epoch AI, "Notable AI Models". https://epoch.ai/data/ai-models. Retrieved 28 October 2026 via AI Safety Tracker, https://aisafetytracker.org/china/.
| Year | Models | Relative magnitude | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2026 | 30 | ||||||||||
| 2025 | 35 | ||||||||||
| 2024 | 31 | ||||||||||
| 2023 | 23 | ||||||||||
| 2022 | 12 | ||||||||||
| 2021 | 21 | ||||||||||
| 2020 | 8 | ||||||||||
| 2019 | 6 | ||||||||||
| 2018 | 1 | ||||||||||
| 2017 | 10 | ||||||||||
Show the remaining 3
|
|||||||||||
| Organisation | Models | Relative magnitude | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Alibaba | 36 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Tsinghua University | 25 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Baidu | 13 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ByteDance | 12 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Z.ai (Zhipu AI) | 12 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| DeepSeek | 10 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Moonshot | 8 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Tencent | 7 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Chinese Academy of Sciences | 6 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Peking University | 6 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Show the remaining 52
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Notability is Epoch AI's editorial judgement, not a census. Organisations are named as Epoch names them, so one company can appear under several names. 2 models are counted in the total and by year but not by organisation, because Epoch's lists of organisations and countries for them do not line up.
Data: Epoch AI, Notable AI Models
Frontier developers
The developers in the Frontier Safety Framework Index with notable models Epoch AI credits to an organisation in China. DeepSeek is based there. A developer is based where most of its credited models were built. Notable AI Models Epoch AI Inclusion is editorial ("notability criteria"), not exhaustive. Compute figures are frequently estimates; check the per-row confidence and notes columns before treating any value as measured. Cite Epoch AI, "Notable AI Models". https://epoch.ai/data/ai-models. Retrieved 18 October 2026 via AI Safety Tracker, https://aisafetytracker.org/china/. Release Safety Disclosure Index This site Our own reading of the system card, model card or safety report each developer published for each release Epoch AI lists from 2025, joined to Epoch's list. Every cell is what a developer says about its own model under its own framework: whether a card came out with the release, whether it reports dangerous-capability results, which government institutes it names as pre-release testers, and whether it says a risk threshold was reached. None of it is verified. A release Epoch adds after the last review is shown as awaiting review. Cite AI Safety Tracker, "Release Safety Disclosure Index". https://aisafetytracker.org/alignment/#releases. Retrieved 18 October 2026. Notable AI Models Epoch AI Inclusion is editorial ("notability criteria"), not exhaustive. Compute figures are frequently estimates; check the per-row confidence and notes columns before treating any value as measured. Cite Epoch AI, "Notable AI Models". https://epoch.ai/data/ai-models. Retrieved 18 October 2026 via AI Safety Tracker, https://aisafetytracker.org/china/.
| Developer | Models credited here | Of all its models | Based in | Card on the day |
|---|---|---|---|---|
| DeepSeek | 10 | 10 | here | 0 of 4 |
| Microsoft | 5 | 47 | United States | 1 of 1 |
Models are credited where Epoch's organisations and countries pair up, so a model whose two lists do not line up is not credited anywhere here. Card on the day counts releases since January 2025 in the Release Safety Disclosure Index, each the developer's own statement.
Data: Epoch AI, Notable AI Models, and the Release Safety Disclosure Index
Not covered
- Depth of the law index. It holds 7 instruments for China and 28 for Australia, which was indexed first and in more depth, so the counts are not a comparison.
- Enterprise use of AI. The enterprise survey on AI adoption is Eurostat's and covers Europe.
- AI datacentres outside Epoch AI's hub. The AI Datacentre Index, which records sites from developers' statements, covers Australia only.
- Official publications. The news feed reads no Chinese government feed.
- Notable models from Hong Kong. Epoch AI records Hong Kong apart from China, and its 15 models are not counted here.
- Electricity, emissions and water at Chinese datacentres. No indexed source publishes them per site or per Chinese grid region.