Jen-Hsun Huang shocked the announcement! Nvidia's market share in China fell from 95% to 0%

NVIDIA CEO Jen-Hsun Huang stated last week that due to the ongoing impact of U.S. export controls, the company's market share in China's advanced AI accelerator market has plummeted from around 95% to 0%. In an interview, Jen-Hsun Huang said, “Currently, we have completely exited the Chinese market, with our market share falling from 95% to 0%.” This is also the first time NVIDIA has publicly quantified the scale of its exit from the Chinese market.

From 95% Dominance to Total Zeroing of the Collapse

(Source: X)

Jen-Hsun Huang did not mention specific products, but his statement clearly points to NVIDIA's data center GPU product line, which seems to have completely lost the Chinese market. Since October 2022, this product line has faced multiple export restrictions. In 2023, U.S. export bans rendered NVIDIA's A800 and H800 chips, designed specifically for the Chinese market, non-compliant; while its new product H20 also encountered its own licensing approval issues.

What does a 95% market share mean? It signifies that NVIDIA is almost in a monopoly position in the advanced AI accelerator market in China. AI laboratories, tech giants, and cloud service providers in China rely almost entirely on NVIDIA's high-end GPUs such as the A100 and H100 when training large language models and executing AI inference. This dependency is not only technical but also ecological. NVIDIA's CUDA platform has become the de facto standard for AI development, with a large number of open-source models and frameworks optimized for CUDA.

The process of falling from 95% to 0% did not happen overnight, but rather went through multiple rounds of tightening policies. In October 2022, the United States implemented export controls on NVIDIA's A100 and H100 for the first time. NVIDIA then launched downgraded versions A800 and H800 in an attempt to bypass the restrictions. In October 2023, the U.S. tightened policies again, and A800 and H800 were also banned. NVIDIA introduced an even lower-spec version, the H20, but the approval process was prolonged, and actual delivery was hindered. After 2024, as the scope of controls expanded to more models, NVIDIA's compliant product options in China have come close to zero.

Regarding this outcome, Jen-Hsun Huang stated, “I can't imagine any policymaker thinking this is a good result—our policies have ultimately led to the United States losing one of the largest markets in the world, with its share dropping to 0%.” Such public questioning of U.S. policies is quite rare among tech CEOs. Most corporate leaders tend to express their dissatisfaction with policies more subtly, while Huang's frank criticism shows his deep disappointment with this result.

Jen-Hsun Huang's argument is logically clear: the purpose of export controls is to delay China's AI development, but the actual effect is to force China to accelerate domestic production, ultimately harming American companies and the technological leadership of the United States. When China no longer relies on NVIDIA, the U.S. will lose not only market share and revenue, but also its influence over China's AI industry and the technological lock-in effect.

Lost 25% revenue, China turns to domestic alternatives

NVIDIA previously disclosed that the Chinese market contributed 20% to 25% of its data center business revenue. In the company's latest financial report, data center business revenue exceeded $41 billion, growing 56% compared to the same period last year. Although this data includes cloud service customers handling various workloads, AI infrastructure remains the company's growth engine. Long-term export restrictions may reshape the demand landscape and supply chain system for this business.

A revenue contribution of 20% to 25% means that NVIDIA earns about 8 to 10 billion dollars annually from the Chinese market. This scale is equivalent to the annual revenue of many medium-sized tech companies. Losing this market is not only a financial loss but also a strategic failure. China is the world's second-largest economy, with the largest pool of engineers and the most active AI startup ecosystem; losing this market means that NVIDIA's influence in the global AI industry chain is weakened.

NVIDIA's cautious expectations are due to the further fragmentation of the AI industry chain. In response to export restrictions, China's large-scale technology companies and AI laboratories have increasingly turned to domestically produced chips or other alternative hardware, accelerating the process of domesticating computing infrastructure. Jen-Hsun Huang pointed out this trend earlier this year, warning that comprehensive restrictions could drive the development of competitive alternative products.

China's Domestic AI Chip Replacement Process

Huawei Ascend Series: Ascend 910B is regarded as a domestic solution comparable to NVIDIA's A100.

Aliyun Hanguang Series: Self-developed chips optimized for cloud AI inference.

Cambrian, Wall Street Technology: Several startups have launched AI training and inference chips.

Software Ecosystem Restructuring: Migrating from CUDA to domestic frameworks such as Ascend and Flying Paddle.

This alternative is not something that can be achieved overnight. The performance of domestically produced chips in China still lags behind that of NVIDIA's most advanced products, but the gap is narrowing. More importantly, once the Chinese AI ecosystem has fully migrated to domestic platforms, even if the U.S. relaxes regulations in the future, it will be very difficult for NVIDIA to regain the market. The lock-in effect of the software ecosystem is stronger than that of hardware; when developers have become accustomed to domestic frameworks and toolchains, the switching costs will be extremely high.

Jen-Hsun Huang stated, “In all of our forecasts… the share of the Chinese market is calculated at 0%. If there is any turnaround in the Chinese market in the future… it will be an additional surprise.” This statement indicates that NVIDIA has given up on short-term expectations for the Chinese market and has completely shifted its strategic focus to other markets. This is also a risk management strategy: by calculating performance guidance at 0%, even if the Chinese market really goes to zero, it will not disappoint investors.

Geopolitics Reshaping the Global AI Industry Chain

As part of the overall strategy to limit China's access to advanced semiconductors, the U.S. government has tightened controls on the export of AI accelerators to China. However, Jen-Hsun Huang's remarks highlight that the actual pace of change in the market landscape is far beyond expectations. The long-term effects of this policy are becoming evident: it has not only changed Nvidia's market landscape but also reshaped the power structure of the global AI industry chain.

From a geopolitical perspective, the U.S. export control strategy is based on the assumption that by restricting China's access to advanced chips, it can delay its AI development and maintain the U.S.'s technological lead. However, Jen-Hsun Huang's warning reveals the potential flaws in this strategy. When restrictions are too stringent, they do not prevent China from developing AI; instead, they force it to invest more resources into developing alternatives. The ultimate outcome could be that U.S. companies lose market share, China gains technological independence, and U.S. technological influence declines.

Despite Jen-Hsun Huang stating that he still hopes Nvidia can return to the Chinese market in the future, the company has effectively excluded the Chinese market from its business expectations. This statement serves both as a reality check for investors and as a subtle pressure on policymakers. As one of the most successful technology companies in the United States, Nvidia's market value has surpassed $500 billion, and the public criticism from its CEO may influence policy debates.

From a broader perspective, Nvidia's experiences in the Chinese market are a microcosm of the larger backdrop of US-China tech competition. Cutting-edge fields such as semiconductors, AI, and quantum computing are undergoing unprecedented geopolitical influences. Companies can no longer make decisions based solely on business logic; they must incorporate geopolitical risks into their strategic considerations. This new normal will have a long-term impact on the global technology industry landscape and innovation pathways.

For NVIDIA, the complete loss of the Chinese market is indeed a significant loss, but the company's overall performance is still showing strong growth. Data center business revenue exceeds $41 billion, growing 56% compared to the same period last year, indicating that demand from the United States, Europe, and other markets is sufficient to offset the absence of the Chinese market. However, in the long run, the strategic cost of losing one of the world's largest markets may gradually become evident in the coming years.

Jen-Hsun Huang's public statement is both an acknowledgment of reality and an expectation for the future. “If there are any turning points in the Chinese market in the future… it will be an additional surprise,” this statement implies that NVIDIA has not completely given up on the Chinese market, but is waiting for changes in the policy environment. This waiting may take several years or even longer, depending on the evolution of US-China relations and the landscape of global technology competition.