Nvidia now holds 34% of China's market for artificial-intelligence chips, according to figures reported by Yahoo Finance, which attributes the decline directly to U.S. export restrictions on advanced semiconductors.

The number marks a sharp contraction for a company that had supplied the overwhelming majority of the accelerators powering China's AI training and inference workloads. Washington's export-control regime, administered over successive rounds of restrictions, has progressively barred Nvidia from selling its highest-performance data-center GPUs to Chinese customers without a licence.

What does the 34% figure mean?

Market share in AI chips is measured by revenue share of accelerators sold into a given market over a defined period. A reading of 34% means that roughly one in three dollars spent on AI chips in China now goes to Nvidia, with the remainder flowing to suppliers outside the scope of U.S. controls — principally domestic Chinese designers and sanctioned-tolerant foreign vendors.

The decline matters on three levels:

• Revenue: China has historically ranked among Nvidia's largest single markets, and each point of share lost translates into orders redirected elsewhere.
• Supply chains: Chinese cloud and AI labs are qualifying alternative accelerators as Nvidia's newest parts remain blocked.
• Policy: The figure gives Washington a quantified indicator that its controls are altering procurement patterns, not merely restricting paper sales.

Each of these effects compounds the longer the restrictions stay in force.

How did the restrictions produce the drop?

The U.S. government has tightened export rules on advanced computing hardware in stages, closing loopholes each time Nvidia introduced a compliant derivative chip for the Chinese market. Every revised rule narrowed the set of products Nvidia could legally ship.

Chinese buyers responded by re-tendering around available alternatives, and domestic vendors gained design wins that would previously have gone to Nvidia by default. The 34% share is the cumulative result of that cycle: successive regulatory walls, successive substitution.

What follows from here?

Nvidia retains the 34% it still holds through the chip variants it can legally export. Whether that residual share holds, erodes further, or stabilises depends on two variables: the next round of U.S. rule-making and the pace at which Chinese fabs and designers close the performance gap.

For now, the number stands as the clearest single measure of how far export policy has already moved the market.

The share loss is quantified; the policy that caused it remains in force.