# China [https://timesof.ai/region/china](https://timesof.ai/region/china) Coverage of Chinese research institutions, national AI initiatives, and regional deployments. China's LLM momentum spans academic labs, state-backed initiatives, and industry deployments focused on domestic infrastructure and sector-specific models. ## Coverage - [Reuters Finds $3.2 Billion of Chinese AI Projects Designed Around Nvidia Chips the US Restricts, a Bet That Now Rests on Washington](https://timesof.ai/2026/11/reuters-3-2b-chinese-ai-projects-built-on-restricted-nvidia-chips) — Reuters found 23B yuan of Chinese AI computing projects planned around restricted Nvidia H100, H200 and H800 chips. The total includes buildings and power. - [Six Chinese Open-Weight Models Carry a Blackwell Fingerprint, an Independent Weight Analysis Finds, but It Cannot Say Where the Chips Were](https://timesof.ai/2026/11/blackwell-fingerprints-deepseek-moonshot-minimax-ant-open-weights-unemployment-labs) — An Unemployment Labs analysis says DeepSeek, Moonshot, MiniMax and Ant Group models were initialized on Nvidia Blackwell GPUs. Location and legality unproven. - [MiniMax's Open-Weight H3-Image Makes 4K Pictures in Seconds, Under a Licence That Shuts Out the US, EU, UK and Korea](https://timesof.ai/2026/11/minimax-h3-image-open-weights-4k-license-excludes-us-eu-uk-korea) — MiniMax's open-weight H3-Image (33.1B DiT, six steps) claims 4K in under 10s. Licence excludes the US, EU, UK and Korea; $1M revenue cap. - [Chinese AI Firms Rush to Lock In Overseas Nvidia Capacity Before Washington Closes the Cloud Loophole, Reuters Reports](https://timesof.ai/2026/11/chinese-ai-firms-rent-overseas-nvidia-capacity-remote-access-reuters) — Reuters: Chinese AI firms are trying to reserve overseas Nvidia compute, fearing allegations around former CEO Jensen Huang speed the Remote Access Security Act. Firms unnamed. - [A Blacklisted Chinese Defence University Bought Nvidia H200s Through a Local Supplier, Reuters Finds in Procurement Records](https://timesof.ai/2026/11/nvidia-h200-northwestern-polytechnical-entity-list-reuters) — Reuters: Entity-listed Northwestern Polytechnical bought two H200s via a local supplier and rented H200 servers, showing export-control gaps. Nvidia did not respond. - [U.S. and China Plan to Compare Pre-Release AI Testing, Bessent Says, While Committing to Nothing That Slows the Race](https://timesof.ai/2026/11/us-china-superintelligence-dialogue-model-testing-bessent) — Bessent says a U.S.-China 'superintelligence' dialogue will cover pre-release model testing and a serious-incident alert channel. No limits on development, no published procedures. - [Alibaba Previews Qwen4-Max for Two Weeks, Subscription-Only and With No Independent Scores Yet](https://timesof.ai/2026/11/alibaba-qwen4-max-preview-1m-context-token-plan-only) — Qwen4-Max preview opens Nov 11 with image/video input and 1M context, via Alibaba's Token Plan only; it retires Nov 25. No independent benchmark listing yet. - [Microsoft Is Lobbying Against a Ban on Chinese Optical Parts That Wire AI Data Centers Together, Reuters Reports](https://timesof.ai/2026/11/microsoft-presses-trump-against-chinese-optical-transceiver-ban-reuters) — Reuters: Microsoft is pressing the Trump administration to drop a planned FCC ban on new Chinese optical transceivers used in AI data centers. Microsoft declined to comment. - [Trump Says He 'Doesn't Mind' AI Guardrails, as Long as They Don't Put the US in 'a Deep Second Position' to China](https://timesof.ai/2026/11/trump-60-minutes-ai-guardrails-not-deep-second-to-china) — Trump told 60 Minutes AI risks are manageable and he doesn't mind guardrails, but not ones leaving the US 'in a deep second position' to China. No proposal yet. - [DeepSeek's Agent App Harness Says It Has 1.54 Million Daily Users, Each Burning About 60 Million Tokens a Day](https://timesof.ai/2026/11/deepseek-harness-1-54m-daily-users-92-trillion-tokens) — DeepSeek says its Harness agent app reached 1.54M daily users in the week to Nov. 4, using about 92 trillion tokens a day. The figures are DeepSeek's, unaudited. - [NYT: U.S. Cleared Nvidia H200s for Chinese Firms With Empty Offices and Ties to Blacklisted Companies](https://timesof.ai/2026/11/us-cleared-nvidia-h200-chinese-firms-empty-offices-nyt) — The NYT says Commerce cleared New Matrix, an empty-office Beijing firm tied to blacklisted Wise Road, to assemble Nvidia H200 servers. BIS says it vets all applications. - [About 600,000 Nvidia H200 Chips Are Stuck in Hong Kong as Beijing Withholds Import Approval, Reuters Reports](https://timesof.ai/2026/11/nvidia-600000-h200-chips-hong-kong-china-withholds-approval) — Reuters says about 600,000 Nvidia H200 chips sit in Hong Kong as Beijing withholds approval for large imports, pushing up grey-market server prices. - [Tencent's Hy4 Adds Vision and Computer Use, and Tencent Says the Open Weights Follow in Days](https://timesof.ai/2026/11/tencent-hy4-vision-computer-use-open-weights-coming) — Tencent's Hy4 adds vision and computer use, trained on Blender and CAD, at preview prices. Weights and system cards are promised in days; scores are Tencent's own. - [Revolut Made a Self-Hosted Chinese Open-Weights Model Its Default for 2,000 Engineers and Says AI Coding Now Costs About a Tenth](https://timesof.ai/2026/11/revolut-self-hosted-glm-default-coding-model-2000-engineers) — Revolut runs Z.ai's GLM-5.3 on 56 B200 GPUs as the default coding model for about 2,000 engineers, cutting AI coding costs about 90%, The Register reports. - [Moffett AI Raises Nearly RMB 1 Billion Again, Five Months After Its Last Round, to Build Sparse-Computing Inference Chips](https://timesof.ai/2026/11/moffett-ai-series-c-plus-nearly-1b-yuan-sparse-inference-chips) — Shenzhen's Moffett AI closed a nearly RMB 1 billion Series C+, its second such round in five months, to develop sparse-computing AI inference chips. - [Meituan's LongCat-2.5 Is a 1.6 Trillion-Parameter Agent Model Trained on Chinese Chips; Its Own Charts Put It Behind Claude Opus 5 on the Hardest Coding Test](https://timesof.ai/2026/11/meituan-longcat-2-5-1-6t-multimodal-domestic-accelerators) — Meituan's LongCat-2.5 has 1.6T parameters, 48B active, trained on Chinese accelerators. Vendor benchmarks trail Claude Opus 5 on Terminal Bench v4.0. - [Kuaishou's Kling AI Closes a 19.8 Billion Yuan Round After Two Investors Fail to Pay, With a GPU Maker and Kingdee Filling the Gap](https://timesof.ai/2026/11/kling-ai-completes-19-8b-yuan-round-kingdee-metax) — Kuaishou's Kling AI completed an RMB19.8B round after two investors failed to pay. Kingdee and GPU maker MetaX took over commitments; Kuaishou keeps control. - [Baidu's Apollo World Simulates 100,000 km of Robotaxi Driving a Day, With Every Other Car and Pedestrian Run as an AI Agent](https://timesof.ai/2026/11/baidu-apollo-world-robotaxi-world-model-100000-km-a-day) — Baidu's Apollo World world model simulates up to 100,000 km of robotaxi driving a day for Apollo Go, with AI-agent traffic and live camera and lidar data. - [Robot Training-Data Startup GenRobot Raises $100 Million From SAIC, Baidu and Galbot Among Others](https://timesof.ai/2026/11/genrobot-ai-100m-series-a-plus-robot-training-data) — Beijing's GenRobot AI raised a $100M Series A+ from SAIC Motor, Baidu, Galbot and others to sell human-demonstration data for training robots. No valuation. - [Meta Built a Humanoid Prototype Called Max With China's Fourier and Started a Second Version in September, The Information Reports](https://timesof.ai/2026/11/meta-fourier-max-humanoid-prototype-the-information) — The Information reports Meta and Fourier built the Max humanoid in about three months and began Max 2 in September. Meta says it is research, not for sale. - [DeepSeek Open-Sources dsh-superagent, a Prototype Coding Agent Built to Spend as Little Human Attention as Possible](https://timesof.ai/2026/11/deepseek-dsh-superagent-open-source-coding-agent-prototype) — DeepSeek's MIT-licensed dsh-superagent works in one GitHub repo, takes requests via GitHub or Feishu and replies with PRs. DeepSeek calls it a research prototype. - [CrowdStrike Says a Second China-Nexus Group, VAPOR PANDA, Used ARTEX and DeepSeek Against More Than 200 Organizations in 11 Days](https://timesof.ai/2026/10/crowdstrike-vapor-panda-artex-deepseek-200-organizations) — CrowdStrike says China-nexus VAPOR PANDA ran ARTEX with DeepSeek against 200+ organizations in 14+ countries in 11 days, separate from Storm-0940. - [Senate Calls Huang and Lutnick to Testify as Bessent Says Commerce Held the Pen on Blacklisting Megaspeed](https://timesof.ai/2026/10/nyt-megaspeed-exec-chinese-intelligence-asset-jensen-huang) — Senate Banking asked Jensen Huang to testify Dec. 3 and Lutnick Dec. 10. Bessent says Commerce has the pen on Megaspeed. DOJ: no spy conclusion yet. - [Ant Group's inclusionAI Releases Ling-3.1-flash-1019, a 560B-Parameter Open-Weight Model Under the MIT License](https://timesof.ai/2026/10/ant-ling-3-1-flash-1019-open-weights-mit) — Ant Group's inclusionAI releases Ling-3.1-flash-1019, a 560B-parameter mixture-of-experts model with 25B active parameters and a 1M-token context, under the MIT license.