The strained relationship between the United States and China has significantly impacted the tech industry, particularly in the field of Artificial Intelligence, where dominance translates to industrial leadership. In recent years, to prevent China from advancing too quickly in AI, the US implemented export restrictions, preventing the purchase of advanced AI chips and machinery. This move has had a profound effect on companies like NVIDIA and their widely used GPUs.
Despite an initial agreement that allowed NVIDIA to sell modified versions of their GPUs in China, it appears to be too late. NVIDIA’s CEO, Jensen Huang, has revealed that due to US policies, the company’s market share for AI GPUs in China is now effectively zero. This situation arose as China has increasingly rejected NVIDIA, opting for domestic solutions.
The US initiated chip bans against China with the explicit goal of hindering its main AI rival, even before generative AI became a market-dominating force. While this strategy might have offered short-term gains for the US, its long-term effectiveness is now in question. China’s rapid advancements, exemplified by the open-source AI model DeepSeek R1 in early 2025—a ChatGPT competitor developed with significantly fewer NVIDIA GPUs and in a shorter timeframe—have put pressure on the US AI industry to accelerate its own progress. In some aspects, such as AI video generation, China may now be surpassing the US.
NVIDIA CEO Reports Zero AI GPU Market Share in China, Blames US Policy
While China continued its relentless progress in AI, NVIDIA’s attempts to supply its GPUs to the Chinese market have been largely unsuccessful. Jensen Huang, NVIDIA’s CEO, has officially declared that the company’s market share for AI GPUs in China is precisely 0%. This is a striking admission, considering NVIDIA’s position as a global leader in AI hardware and graphics cards. Huang attributes this zero market share to US export policies, stating that Chinese customers have lost interest in NVIDIA’s graphics cards, preferring to utilize their own domestically produced chips.
Previous analyst estimates had projected NVIDIA’s AI market share in China to fall from 66% in 2024 to 8% in the coming years. However, Huang’s recent statement indicates the reality is even more severe. It’s important to note that this 0% refers to direct sales from NVIDIA to clients in China. Interest in NVIDIA’s GPUs still exists, with some companies reportedly acquiring them through third-party countries to circumvent restrictions. Jensen Huang believes the US measures have been excessively restrictive and ultimately counterproductive.
China Rejects Expensive NVIDIA GPUs in Favor of Local Chips and Accelerators Like Huawei’s
The US began implementing restrictions in 2022, initially blocking the sale of NVIDIA A100 and H100 GPUs. In response, NVIDIA developed reduced-performance versions, the A800 and H800. These measures were further tightened in 2024, extending to even the modified NVIDIA H20 graphics cards. By 2025, instances of GPU smuggling increased as China sought to acquire hardware through third-party markets to bypass controls.
In early 2026, the US permitted chip sales once more, but with stringent conditions. NVIDIA could sell H200 chips to China only if a 25% tariff was applied, in addition to all price increases. This ultimatum proved to be a breaking point, as Chinese companies outright rejected the idea of purchasing such expensive graphics cards. Consequently, they fully shifted to using local chips, including AI accelerators like Huawei’s Ascend. With the near-complete dominance of “Made in China” AI chips, interest in American hardware has dwindled as China pursues its goal of technological independence.
