Nvidia’s China Market Share Hits Zero as Export Controls Backfire
Jensen Huang's admission that US chip restrictions eliminated his company's entire Chinese AI accelerator business validates the policy's effectiveness—and its strategic cost.
Nvidia CEO Jensen Huang disclosed on 30 April that the company’s market share in China’s AI accelerator market has collapsed to zero, down from 95% two years ago, marking the most concrete evidence yet that US export controls are reshaping the global semiconductor landscape.
The admission, made during an interview with the Special Competitive Studies Project, represents a strategic inflection point in US-China tech decoupling. China previously accounted for $19.67 billion in Nvidia revenue in fiscal 2026—roughly 20-25% of the company’s data center business. That revenue stream has now been eliminated entirely, with Nvidia guiding for zero China contribution in its current fiscal quarter.
Export Controls Achieved Their Primary Objective
The market share wipeout validates the technical effectiveness of the Biden administration’s semiconductor export restrictions, which began in October 2022 with bans on Nvidia’s A100 and H100 chips and tightened progressively through 2026. The controls forced Nvidia to develop China-specific products like the H20—a downgraded chip designed to comply with compute density thresholds—but even those workarounds became untenable after April 2025, when licensing requirements effectively blocked shipments.
Nvidia took a $4.5 billion charge in its most recent quarter tied to H20 excess inventory and purchase obligations, according to Computer Weekly. The financial hit underscores how abruptly the policy cut off what had been a reliable revenue stream. Bernstein analysts estimate Nvidia’s China AI GPU market share will stabilise around 8% in coming years—a structural compression from the 66% share the company held in 2024.
Huawei and Domestic Competitors Fill the Vacuum
The policy’s secondary effect—accelerating Chinese self-sufficiency—is now fully materialised. Tom’s Hardware reports Huawei’s AI chip revenue is forecast to reach $12 billion in 2026, up 60% from $7.5 billion in 2025, based on confirmed orders for its Ascend 950PR accelerator. Domestic competitors including Cambricon, Moore Threads, and MetaX are ramping production to serve a market that no longer has access to Nvidia’s flagship products.
“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.”
— Jensen Huang, CEO, Nvidia
Huang’s critique centres on the strategic calculus of surrendering a market projected to reach $50 billion annually. “Maybe it made sense at the time, but I think the policy really needs to be dynamic and needs to stay with the times,” he told the Special Competitive Studies Project. The comments reflect a corporate position that the controls achieved short-term disruption but created long-term competitive threats by forcing China to build indigenous capabilities that will eventually compete globally.
Margin Pressure and Geographic Concentration Risk
Nvidia posted $215.9 billion in total revenue for fiscal 2026, with data center products accounting for $193 billion—89.6% of the total, per TechLoy. Despite the China loss, the company is guiding for 77% revenue growth in the current quarter, indicating hyperscaler demand in the US and Europe is more than offsetting the shortfall. But the geographic concentration creates new vulnerabilities: Nvidia is now dependent on a narrower customer base, primarily US cloud providers and European AI labs, for growth that previously had more diversified geographic sources.
CFO Colette Kress warned in February that “losing access to the China AI accelerator market, which we believe will grow to nearly $50bn, would have a material adverse impact on our business going forward and benefit our foreign competitors in China and worldwide,” according to Computer Weekly. The statement proved prescient—Huawei’s 60% revenue surge this year demonstrates how quickly domestic players captured the void Nvidia left behind.
China’s Self-Sufficiency Threshold
Analysis from War on the Rocks suggests China does not need to match TSMC’s 3nm process leadership to achieve strategic autonomy in AI—it only needs chips capable of training commercial large language models at scale. By that standard, China’s progress is significant. Huawei’s Ascend chips, manufactured on mature nodes by SMIC, are proving sufficient for domestic AI development, even if they lag Nvidia’s flagship products on raw performance.
Huang noted that China possesses structural advantages in AI development that extend beyond chip access: “They have cheaper energy. They have incredible talent. The number of AI researchers in China is quite extraordinary, it’s one of their national treasures.” The comment underscores that export controls addressed only one dimension of China’s AI capabilities—access to cutting-edge silicon—while leaving intact advantages in talent, capital, and state coordination.
Supply chain analysis from Oplexa maps out China’s expanding foundry capacity, with SMIC ramping to 45,000-60,000 wafers per month through 2026. The buildout is focused on mature nodes rather than bleeding-edge processes, reflecting a strategic choice to prioritise self-reliance over performance leadership. For commercial AI applications—chatbots, recommendation engines, video synthesis—the performance gap is manageable.
What to Watch
Nvidia’s next earnings call will clarify whether hyperscaler demand can sustain 77% growth rates without China. Watch for guidance revisions if AWS, Microsoft, or Google slow their infrastructure spending. On the policy front, the Trump administration has signalled potential flexibility on export controls, but any loosening would likely come too late to restore Nvidia’s position—Chinese buyers have already switched to Huawei and domestic alternatives with multi-year contracts in place. The more consequential development is whether Chinese AI chips begin appearing in third-country markets, competing with Nvidia globally rather than just domestically. If Huawei’s Ascend chips start winning design wins in Southeast Asia or the Middle East, the strategic cost of the export controls will compound beyond the immediate revenue loss Huang disclosed.