Nvidia's dominance in China's artificial intelligence chip market has completely evaporated, and the company's own CEO is now questioning whether the US export restrictions that caused the collapse actually serve American interests. Jensen Huang stated that Nvidia's market share in China has dropped to zero percent, a stunning reversal from just two years ago when the company controlled the lion's share of China's AI accelerator market.

How Are Chinese Companies Filling the Void Left by Nvidia?

With American chips no longer available, Chinese technology companies have rapidly moved to capture the market that Nvidia once dominated. The shift has been dramatic and swift, with domestic vendors now positioned to cover up to 80 percent of demand. This includes companies developing both their own silicon and software alternatives to replace Nvidia's technology stack.

• Hardware Development: Chinese firms like Huawei, Cambricon, Moore Threads, and MetaX are advancing domestically produced silicon to replace restricted US chips.
• Cost-Efficiency Strategy: SenseTime, a sanctioned Chinese AI pioneer, is shifting focus toward multimodal AI models that run on cheaper hardware, claiming it can reduce cost-per-query by as much as 70 percent.
• Legacy Hardware Support: New Chinese AI models are optimized to run on older-generation GPUs and domestic chips, making them accessible to developing nations and small-to-medium enterprises that cannot afford premium US solutions.

Is the Export Control Strategy Actually Backfiring?

Huang expressed serious concerns about the strategic wisdom of the export restrictions. "Conceding an entire market the size of China probably does not make a lot of strategic sense, so I think that has already largely backfired," he stated. The executive argued that while the policy may have made sense when first implemented, circumstances have changed dramatically, and the restrictions may now be accelerating the very outcome they were designed to prevent: Chinese self-sufficiency in AI technology.

Huang

"In China, we have now dropped to zero. Conceding an entire market the size of China probably does not make a lot of strategic sense, so I think that has already largely backfired. Maybe it made sense at the time, but I think the policy really needs to be dynamic and needs to stay with the times," said Jensen Huang.

Jensen Huang, CEO at Nvidia

Huang emphasized that American companies maintaining a presence in China would actually extend the global reach of the American AI technology stack, particularly in software. The CUDA programming framework, which Nvidia developed, remains a significant advantage that Chinese companies have not yet fully replicated. However, with Nvidia completely shut out of the market, even this advantage may erode over time as Chinese developers build alternative software ecosystems.

What Makes Chinese AI Models Competitive Despite Export Restrictions?

The conventional wisdom in Silicon Valley has been that raw computing power and massive training budgets determine AI leadership. Chinese companies are proving this assumption wrong by focusing on algorithmic efficiency instead. SenseTime's new "Nova-6" model demonstrates this approach, achieving comparable performance to Western models while requiring a fraction of the computing resources.

The cost differences are striking. SenseTime's model required $12 million to train, compared to over $100 million for a GPT-4 equivalent. Inference costs, which determine the price users pay per query, are similarly dramatic: SenseTime charges approximately 0.05 Kenyan shillings per 1,000 tokens, compared to 0.25 shillings for average US models. For developing nations and price-sensitive markets, this represents a game-changing advantage.

SenseTime's co-founder Lin Dahua reframed the entire competition. "The 'AI race' is often misunderstood as a race for the fastest car. In the real world, the winner is the one who builds the most reliable and affordable bus," he explained. This philosophy is driving SenseTime's aggressive expansion into emerging markets across Southeast Asia, the Middle East, and Africa, where Western AI services are often prohibitively expensive.

Why Does China Have Structural Advantages Beyond Just Chips?

Huang acknowledged that even without access to leading American AI chips, China remains a formidable competitor due to factors that export controls cannot restrict. These structural advantages include cheaper energy costs, exceptional talent in mathematics and science, and an extraordinarily large pool of AI researchers that Huang described as "one of their national treasures".

Huang

The combination of these advantages means that restricting chip exports may slow Chinese AI development in the short term, but it cannot prevent it indefinitely. Instead, the restrictions may simply accelerate China's investment in domestic alternatives and push the country toward complete technological self-sufficiency. This outcome could ultimately weaken American influence over global AI standards and governance frameworks.

Huang warned that fear-driven export policies could have broader consequences for global AI deployment. "Fear-driven narratives and export controls could slow the deployment of AI more broadly as China and other regions embrace it more aggressively as an economic tool," he noted. Long-term American leadership, he argued, depends less on restricting rivals and more on ensuring that the American AI ecosystem dominates globally through superior technology and innovation.

Huang

What Does This Mean for Global AI Governance?

As Chinese AI companies expand into developing markets with cheaper, efficient models, they are simultaneously establishing different standards for AI governance and ethics. SenseTime's "Digital Silk Road" strategy is being supported by the Chinese government as a way to set global AI standards that differ from Western liberal models. This could create a fragmented global AI landscape with two competing ecosystems operating under entirely different technical and ethical frameworks.

For countries in the Global South, the choice is increasingly clear: adopt expensive Western AI services or embrace cheaper Chinese alternatives. The economic pressure to choose the latter is powerful, especially in regions where currency fluctuations and limited budgets make US dollar-priced services unaffordable. SenseTime's partners are already offering pricing in local currencies like Kenyan shillings, removing a major barrier to adoption.

The irony is stark. By attempting to slow Chinese AI development through export restrictions, US policymakers may have inadvertently accelerated the emergence of a Chinese-led alternative to the American-dominated global AI ecosystem. Whether this outcome serves long-term American strategic interests remains hotly debated among policymakers and technology leaders.