The numbers that emerged this week from IDC’s latest market analysis mark a watershed moment in the global AI chip race. For the first time, Chinese semiconductor companies have collectively captured more than 40% of their home market for AI accelerators, a milestone that would have seemed implausible just two years ago, when Nvidia’s dominance was near-total and domestic alternatives were largely confined to research labs and government pilots.
According to IDC data reported by Reuters, Chinese chipmakers delivered roughly 1.65 million AI GPUs into the domestic market during 2025, pushing their combined share to approximately 41%. Nvidia’s slice of the same market has fallen to below 60%, a dramatic reversal for a company that once supplied the overwhelming majority of the AI compute powering China’s technology sector.
The catalyst for this shift is well understood: successive rounds of US export controls, tightened most recently in late 2024, have progressively cut Chinese buyers off from Nvidia’s most capable hardware. The H100 and H200 series were restricted first; subsequent rules targeted the downgraded H20 chip that Nvidia had designed specifically to comply with earlier controls. Each restriction created a new opening for domestic suppliers.
Huawei has been the most prominent beneficiary. Its Ascend 910B and 910C AI accelerators have been adopted by a growing number of Chinese cloud providers and state-owned enterprises, driven in part by direct government pressure on data centers to source hardware domestically. Smaller players, including Cambricon, Biren Technology, and Moore Threads, have also expanded their customer bases, though at a more modest scale.
The performance gap between Chinese chips and Nvidia’s best products remains real. On training workloads for frontier-scale models, domestic hardware still lags considerably. But for inference, the task of running a trained model to generate outputs, the gap is narrower, and for many commercial applications the difference is manageable. As Chinese AI companies increasingly focus on deploying rather than training models, the practical relevance of that remaining gap diminishes.
The financial implications for Nvidia are significant. China accounted for roughly 13% of Nvidia’s total revenue in its most recent fiscal year, down from over 20% before the first round of export controls. Further erosion of its market share in a country that remains one of the world’s largest consumers of AI compute represents a meaningful long-term headwind, even as the company continues to post record revenues driven by demand elsewhere.
For China’s semiconductor ecosystem, the milestone carries symbolic as well as commercial weight. The country’s leadership has made chip self-sufficiency a central pillar of its industrial strategy, and the 15th Five-Year Plan, adopted earlier this year, sets explicit targets for domestic supply to cover 80% of the country’s semiconductor needs by 2030. The AI chip market, where domestic players have made the fastest progress, is the most visible proof point for that ambition.
The trajectory also has implications for the broader US-China technology competition. Washington’s export control strategy was premised, at least in part, on the assumption that restricting access to advanced chips would slow China’s AI development. The IDC data suggests the effect has been more complex: restrictions have accelerated domestic chip development and procurement, even as they have imposed real costs and delays. Whether the net effect has been to slow China’s AI progress, or to hasten the creation of a parallel and increasingly self-sufficient supply chain, is a question that policymakers on both sides are now actively debating.
What is not in dispute is the direction of travel. Chinese chipmakers have gone from supplying roughly one-quarter of their home market to more than two-fifths in the space of a single year. If that pace of share gain continues, and there is every reason to expect it will, given the policy tailwinds and the investment flowing into the sector, Nvidia’s position in China will look very different again by the time the next set of annual figures arrives.