China’s top artificial intelligence startup, DeepSeek, has received conditional approval to purchase Nvidia’s H200 AI chips. This decision, confirmed by two sources familiar with the matter, comes as the Chinese government finalizes regulatory conditions that will govern the transaction.
According to reports, major companies including ByteDance, Alibaba, and Tencent have also been given permission to acquire over 400,000 H200 chips collectively. This development highlights the growing demand for advanced AI technology in China, despite ongoing tensions between the United States and China regarding technology exports.
During a press briefing in Taipei, Nvidia CEO Jensen Huang expressed that his company had not yet received definitive information about the approvals. He indicated that the licensing process is still under negotiation. Nvidia has not commented specifically on DeepSeek’s approval, leaving some uncertainty in the air regarding the final terms of the deal.
The regulatory approvals have been granted by China’s ministries of industry and commerce, but the terms are being finalized by the National Development and Reform Commission (NDRC). Requests for comments from the relevant Chinese ministries went unanswered, adding to the ambiguity surrounding the approval process.
DeepSeek has made headlines since early last year with AI models that significantly undercut the costs of similar products from American rivals like OpenAI. The H200 chip, which stands as Nvidia’s second most powerful AI chip, has become a contentious point in US-China relations. While there is strong demand from Chinese firms and recent US approvals for exports, Beijing’s reluctance to permit imports has posed a significant barrier.
Earlier this month, the US officially authorized Nvidia to sell the H200 to China, responding to the robust interest in the technology. Nonetheless, Chinese authorities retain the ultimate authority over whether shipments will be allowed into the country. Any potential purchases of H200 chips by DeepSeek could attract scrutiny from US lawmakers, particularly following allegations that Nvidia assisted DeepSeek in developing AI models utilized by the Chinese military.
In addition to the chip acquisition, DeepSeek is set to launch its next-generation AI model, dubbed V4, which is expected to feature enhanced coding capabilities. The anticipated launch date is in mid-February 2024, according to a report by The Information.
As the global tech landscape evolves, the implications of these transactions will likely resonate beyond the immediate business interests, potentially influencing broader geopolitical dynamics. The situation underscores the delicate balance between technological advancement and regulatory oversight as nations navigate the complexities of the AI sector.
