Key Moments
- Beijing has granted Nvidia approval to sell its H200 AI chips in China after months of regulatory uncertainty.
- Nvidia is developing a version of the Groq AI chip for use in China, with availability expected in May, according to a source cited by Reuters.
- CEO Jensen Huang said Nvidia has licenses and purchase orders from “many customers in China” for the H200, allowing production to restart.
Regulatory Breakthrough for Nvidia’s H200 in China
Nvidia has obtained approval from Chinese authorities to sell its H200 artificial intelligence chip in China, according to people familiar with the situation. The decision marks a significant step for the U.S. semiconductor producer as it navigates complex regulatory requirements in both Washington and Beijing.
The H200 has become a focal point in U.S.-China technology tensions, and the Chinese market previously accounted for 13% of Nvidia’s total revenue. Despite strong interest from Chinese customers and authorization from U.S. regulators to export the chip, Beijing’s reluctance to clear imports had effectively blocked H200 shipments into the country.
Production Restart Backed by New Licenses
Earlier on Tuesday, Nvidia CEO Jensen Huang said the company now holds licenses for “many customers in China” for the H200 and has received purchase orders from “many” companies, enabling the firm to restart production of the chip.
“Our supply chain is getting fired up,” Huang said at a press conference.
Nvidia had previously halted manufacturing of the H200 last year due to escalating regulatory obstacles in both the U.S. and China, according to a prior report referenced in the article. The company had been waiting for months for authorization from both governments. It has already received certain U.S. licenses, and a source familiar with the matter said Beijing has now also granted licenses covering many Chinese customers.
A spokesperson for the Chinese embassy in Washington stated they were “not aware of the specifics,” and referred inquiries to “the competent authorities.” CNBC reported on Tuesday that Huang said Nvidia has now obtained clearance from both sides.
A representative from a Chinese company said they were not sure whether the Chinese government had issued final approval, but that Nvidia had informed them they could proceed to submit purchase orders.
Regulatory Filings and Earlier Approvals
In a filing with the U.S. Securities and Exchange Commission late last month, Nvidia disclosed that in February U.S. authorities granted a license permitting “small amounts of H200 products to specific China-based customers.”
In January, Reuters reported that China had provided preliminary approval to three of its major technology firms – ByteDance, Tencent and Alibaba – as well as AI startup DeepSeek to import the H200, although conditions for final regulatory clearance in China were still being determined at that time. The Chinese companies did not immediately respond to emailed requests for comment.
China AI Stocks React to Huang’s OpenClaw Remarks
Huang’s upbeat comments about the AI agent OpenClaw, which has seen rapid uptake in China, helped push certain Chinese AI-related stocks to record levels on Wednesday. Shares in large-language model developers MiniMax and Zhipu AI jumped more than 19% each after Huang said OpenClaw was “definitely the next ChatGPT.”
Nvidia Positions Groq Chip for China Inference Market
Alongside the H200 progress, Nvidia is also working on a version of the Groq AI chip intended for the Chinese market, Reuters reported earlier on Tuesday, citing two people familiar with the matter.
Nvidia plans to use Groq chips for inference workloads, where AI systems respond to queries, generate code, or perform tasks for users. In products demonstrated this week, Nvidia intends to pair its upcoming Vera Rubin chips – which cannot be sold in China – with Groq chips.
Although Nvidia is the dominant player in AI training hardware, the company faces significantly stronger competition in the inference segment. Several major Chinese technology firms, including AI-focused groups such as Baidu, already design and manufacture their own inference chips.
Groq Variant and Market Strategy
According to one of the sources cited by Reuters, the Groq-based chips being prepared for China are not cut-down versions and are not being built solely for that market. Instead, the new variant can be integrated with other systems, the source said, adding that the Groq chip is expected to become available in May.
Nvidia did not immediately respond to a request for comment.
Regulatory and Product Developments at a Glance
| Item | Details |
|---|---|
| H200 China approval status | Beijing has granted licenses for many Chinese customers, according to sources familiar with the matter |
| U.S. export license for H200 | February license allows “small amounts of H200 products to specific China-based customers” |
| Chinese firms with preliminary import approval (earlier report) | ByteDance, Tencent, Alibaba, and AI startup DeepSeek |
| Groq chip plan for China | Nvidia preparing a variant that can be sold in China for AI inference |
| Expected timing for Groq chip availability | May, according to a source cited by Reuters |





