Beijing’s sweeping directive marks one of its boldest moves yet to eliminate foreign technology from critical infrastructure — a blow to Nvidia, and a boost for Huawei and China’s AI chip ecosystem.
In a decisive escalation of its tech sovereignty ambitions, China has reportedly ordered that all state-funded data centre projects use only domestically-produced artificial intelligence (AI) chips.
According to sources cited by local media, regulatory authorities have instructed that any new data centre projects receiving public funding must deploy exclusively Chinese-made AI processors. Projects still under construction — and less than 30% complete — have been told to remove or cancel orders for foreign chips, while those at a more advanced stage will be reviewed on a case-by-case basis.
The guidance marks one of Beijing’s strongest attempts yet to purge foreign hardware from its digital infrastructure, underscoring its determination to accelerate AI chip self-sufficiency even as trade tensions with Washington remain temporarily paused.
Strategic pivot amid US sanctions
Access to advanced AI chips — especially Nvidia’s high-end GPUs — has long been a geopolitical flashpoint in US-China relations. The latest directive comes shortly after US President Donald Trump said Washington would “let them deal with Nvidia but not in terms of the most advanced” chips following talks with President Xi Jinping.
China’s new order could further diminish Nvidia’s already shrinking presence in the country. The US company’s share of China’s AI chip market has plummeted from 95% in 2022 to nearly zero, according to Nvidia’s disclosures.
The ban reportedly covers even Nvidia’s H20 processors — the downgraded chips still permitted for sale under US export rules — as well as high-end models like the H200 and B200 that are officially restricted but occasionally reach Chinese buyers through grey-market routes.
Huawei, local firms poised to gain
Analysts say the move could accelerate growth for domestic semiconductor leaders such as Huawei Technologies, Cambricon, MetaX, Moore Threads, and Enflame, all of which have been developing competitive AI chips amid US curbs on advanced semiconductor equipment.
Still, China faces a steep challenge: while US firms like Microsoft, Meta, and OpenAI are investing hundreds of billions of dollars into data centres powered by Nvidia’s latest processors, Chinese manufacturers continue to struggle with production bottlenecks and yield limitations.
The directive’s impact is already visible. Several large-scale projects — including one in a northwestern province that had planned to use Nvidia chips — have reportedly been halted pending compliance reviews. Since 2021, more than $100 billion in state funding has flowed into AI data centre infrastructure across China.
Balancing act for Beijing
While Beijing’s order is designed to speed up the nation’s journey toward chip independence, it also risks widening the technological gap with the US in AI computing power. Experts warn that self-reliance, though strategically vital, could slow short-term progress in large-scale model training and cloud services.
Despite these hurdles, major Chinese tech firms — including Alibaba and DeepSeek — have rolled out competitive AI models trained on homegrown chips, signalling early success in Beijing’s push to localize critical technologies.
Nvidia turns to India
With its China business under growing strain, Nvidia is pivoting to India. The company has joined the India Deep Tech Alliance, a consortium of Indian and US investors backing deep-tech startups in South Asia.
According to Reuters, the alliance — launched in September 2025 with an initial $1 billion fund — has since secured $850 million in new commitments. Members include Qualcomm Ventures, Activate AI, InfoEdge Ventures, Chiratae Ventures, and Kalaari Capital.
As part of the partnership, Nvidia will provide technical guidance and policy inputs to Indian startups working in sectors such as AI, robotics, space, and semiconductors. The move aligns with India’s $12 billion national program to boost R&D and deep-tech capacity, with startup funding in the segment surging 78% last year to $1.6 billion, according to Nasscom.
