🚨 China Greenlights Nvidia’s Power Chip for DeepSeek—But With Strings Attached, as AI Geopolitics Hit a New Flashpoint

China has quietly taken a decisive — and controversial — step in the global AI arms race.

According to people familiar with the matter, Beijing has approved its top AI startup, DeepSeek, to purchase Nvidia’s powerful H200 artificial intelligence chips, marking a major breakthrough for China’s AI ambitions — but one wrapped in tight regulatory conditions that are still being finalized.

The move places DeepSeek alongside tech giants ByteDance, Alibaba, and Tencent, which Reuters reported earlier this week had collectively received permission to buy more than 400,000 H200 chips. Together, the approvals signal China’s growing urgency to secure cutting-edge AI hardware — even as geopolitical tensions with Washington remain high.

Nvidia’s H200, its second-most powerful AI chip, has become a lightning rod in U.S.–China tech relations. The chip is essential for training and deploying large-scale AI models, and demand from Chinese firms has surged as competition in generative AI intensifies.

While the U.S. formally cleared exports of the H200 to China earlier this month, Beijing still holds the final veto. Until now, Chinese regulatory hesitation — not U.S. restrictions — has been the main bottleneck holding up shipments.

Sources say China’s Ministry of Industry and Information Technology and Ministry of Commerce have granted approvals for DeepSeek and the other firms, but with conditions being crafted by the powerful National Development and Reform Commission (NDRC) — a sign that Beijing is weighing national strategy as much as commercial demand.

Despite the reports, Nvidia CEO Jensen Huang struck a cautious note while speaking to reporters in Taipei.

Nvidia declined to comment specifically on DeepSeek’s approval, highlighting the opacity surrounding the negotiations — and how sensitive the issue has become.

DeepSeek isn’t just another startup.

The company shook the global tech sector last year after unveiling AI models that reportedly cost a fraction of what U.S. rivals like OpenAI spend, raising eyebrows across Silicon Valley and Washington alike.

Now, with access to Nvidia’s H200 potentially within reach, DeepSeek could supercharge its ambitions. The company is expected to launch its next-generation AI model, V4, in mid-February, featuring advanced coding capabilities that could challenge established Western players.

Any H200 purchases by DeepSeek are almost guaranteed to trigger scrutiny in the U.S. Congress.

Just days ago, Reuters reported that a senior U.S. lawmaker accused Nvidia of helping DeepSeek refine AI models later used by the Chinese military, according to a letter sent to U.S. Commerce Secretary Howard Lutnick. While the claims remain unproven, they underscore how quickly commercial AI deals can morph into national security flashpoints.

China’s cautious approval — greenlighting purchases but layering in regulatory safeguards — reflects a delicate balancing act:

For Nvidia, the situation highlights both the massive opportunity and mounting risk of doing business in China — one of its most important AI markets.

For the world, it’s another reminder that the future of artificial intelligence isn’t just being built in labs — it’s being negotiated in ministries, parliaments, and geopolitical backrooms.

And as DeepSeek prepares its next big AI leap, all eyes are now on whether those final regulatory conditions will unlock a new chapter — or spark the next global tech confrontation.