China to let major AI companies buy Nvidia H200 chips, The Information reports
For the first time in a long while, China will allow its largest domestic AI companies to buy Nvidia H200 chips, in limited volumes and with government approval. The Information reported the news, citing sources. If confirmed, the move would be a notable sign that US technology restrictions on selling advanced AI chips to China are easing.
AI-processed from Bloomberg Tech; edited by Hamidun News
For the first time in several years, China will allow major AI companies to buy Nvidia H200 chips, though only in limited volumes and with government approval. The Information reported the news, citing sources familiar with the matter; Bloomberg Tech relayed the report.
What is the Nvidia H200
The H200 is one of Nvidia's most powerful accelerators for training and running inference on large language models, and the successor to the H100. Its main upgrade is 141 GB of HBM3e memory and significantly higher memory bandwidth than its predecessor. Both matter when working with modern LLMs that have billions of parameters.
- The H200 builds on the Hopper architecture, a key generation of Nvidia AI chips
- 141 GB of HBM3e memory is a significant jump from the H100's 80 GB
- The chip falls into a product category subject to US export controls
- Only "top" Chinese AI companies will be able to buy it, and only in limited quantities
Why China was barred from buying these chips
Since October 2022, the US Department of Commerce has restricted exports of Nvidia's most powerful AI accelerators, including the A100, H100, and H200, to China, citing the risk that the technology could be used for military and intelligence purposes. Nvidia had to develop cut-down versions of its chips specifically for the Chinese market, including the H20. Access to that model has also repeatedly opened and closed amid trade negotiations between Washington and Beijing.
What it means
If The Information's report is confirmed, it would be one of the most visible signs yet that the US-China tech standoff over AI compute is easing. Chinese companies would gain access to more powerful hardware for training models, and Nvidia would regain part of the market it lost to sanctions.
- Nov 1, 2026Local models lose to Opus and Codex: 0.21 on a MacBook versus 1.00 for Codex
- Nov 1, 2026Habr author fine-tunes YOLOv8l on dental X-rays and pushes mAP50 to 0.642
- Nov 1, 2026Voice bots built on LLMs turn to Jev, a fast classifier, for decisions in 200–500 ms
- Nov 1, 2026MoneyGate safety layer blocked all 42 fraudulent requests to local AI models
Key Questions Answered
What is the Nvidia H200?
The H200 is one of Nvidia's most powerful accelerators for training and running inference on large language models, and the successor to the H100. Its main upgrade is 141 GB of HBM3e memory and significantly higher memory bandwidth than its predecessor. Both matter when working with modern LLMs that have billions of parameters. The H200 builds on the Hopper architecture, a key generation of Nvidia AI chips 141 GB of HBM3e memory is a significant jump from the H100's 80 GB The chip falls into a product category subject to US export controls Only "top" Chinese AI companies wi
Why China was barred from buying these chips?
Since October 2022, the US Department of Commerce has restricted exports of Nvidia's most powerful AI accelerators, including the A100, H100, and H200, to China, citing the risk that the technology could be used for military and intelligence purposes. Nvidia had to develop cut-down versions of its chips specifically for the Chinese market, including the H20. Access to that model has also repeatedly opened and closed amid trade negotiations between Washington and Beijing.
What it means?
If The Information's report is confirmed, it would be one of the most visible signs yet that the US-China tech standoff over AI compute is easing. Chinese companies would gain access to more powerful hardware for training models, and Nvidia would regain part of the market it lost to sanctions.
Test Frontier LLMs & Autonomous Agents in One Workspace
Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.