NVIDIA’s Zero Market Share in China Marks Point of No Return in Tech Decoupling
Jensen Huang's confirmation that US export controls drove NVIDIA from 95% to 0% market share validates containment strategy—but may have irreversibly accelerated China's chip self-sufficiency agenda.
NVIDIA CEO Jensen Huang confirmed the company has lost its entire China market share—dropping from 95% to zero—due to US export controls on advanced AI chips, marking a definitive inflection point in tech decoupling that may have permanently altered global AI competition dynamics.
The collapse, driven by Biden-era restrictions on H100 and H200 chip exports, represents an estimated 15-20% revenue loss for NVIDIA but validates the narrow objective of US semiconductor containment policy. The unintended consequence: China’s forced pivot to domestic chip development has accelerated rather than stalled, with Huawei, Alibaba, and Baidu now locking in indigenous supply chains that may render future US leverage obsolete.
“We went from 95% market share to 0%, and so I can’t imagine any policymaker thinking that that’s a good idea,” Huang stated in a May 2026 interview, according to Tom’s Hardware. “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.”
95%
0%
$4.5B
20-25%
The Export Control Escalation
The Biden administration initiated export restrictions on NVIDIA’s most advanced AI accelerators in October 2022, targeting H100 and later H200 chips capable of training frontier AI models. NVIDIA recognized $4.5 billion in charges tied to these restrictions in Q1 fiscal 2026, per Let’s Data Science. China had previously represented 20-25% of NVIDIA’s data center revenue, a segment that generated over $41 billion in the company’s most recent results.
The Trump administration initially tightened restrictions in April 2025, then reversed course in December 2025 by granting limited H200 export licenses—contingent on a 25% revenue share flowing to the US Treasury. But the policy reversal came too late. China blocked H200 imports anyway, and NVIDIA CFO Colette Kress confirmed in February 2026 that the company generated zero revenue from China datacenter operations despite the licenses, according to The Register.
“While small amounts of H200 products for China based customers were approved by the US government, we have yet to generate any revenue, and we do not know whether any imports will be allowed into China,” Kress told investors.
China’s Domestic Chip Breakout
The export vacuum triggered an acceleration in Chinese domestic chip development that may have permanently shifted competitive dynamics. Huawei’s Ascend 950PR, targeting 750,000 unit shipments in 2026 with mass production beginning in April, has become the anchor of China’s AI infrastructure buildout, per Technology.org. ByteDance, Alibaba, and Tencent have placed significant orders following the late April release of DeepSeek V4—an AI model optimized specifically for Huawei’s Ascend architecture.
“This is a big deal for China’s AI industry. It’s a good thing for the entire domestic AI industry,” said He Hui, director of semiconductor research at Omdia, commenting on DeepSeek V4’s Huawei optimization to Asia Financial.
“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.”
— Jensen Huang, NVIDIA CEO
Performance benchmarks show Huawei’s Ascend 910C achieves roughly 60% of NVIDIA’s H100 capability in real-world tasks, according to DeepSeek research cited by the Council on Foreign Relations. While the H200 remains superior, supply is blocked by Beijing’s import ban—rendering the performance gap strategically irrelevant.
Alibaba’s proprietary PPU chip costs 40% less than NVIDIA’s export-compliant H20, while Baidu’s Kunlun secured a ¥1 billion order from China Mobile in September 2025, data from 36Kr shows. Domestic AI Chips captured 30% market share in China during 2024, shipping over 820,000 units as NVIDIA’s share dropped from 85% to 70% that year, per IDC data.
Foundry Consolidation and Supply Chain Lock-In
China is consolidating its semiconductor manufacturing base to lock in self-sufficiency gains. SMIC is acquiring its Beijing subsidiary SMIC Jingcheng for ¥40 billion (~$5.8 billion), while Hua Hong is purchasing Shanghai Huali for $1.2 billion—deals announced in January 2026 and driven explicitly by export control pressure, Tom’s Hardware reported. Chinese foundries are expected to control over 25% of global mature-node capacity by end of 2025, expanding aggressively at 28nm and 22nm nodes.
China’s semiconductor self-sufficiency target stands at 50% by 2025 according to TrendForce, with Huawei founder Ren Zhengfei targeting 70% by 2028. Beijing is backing this push with $70 billion in subsidies for domestic chipmaking, though US equipment bans—particularly on EUV lithography and advanced deposition tools—constrain Chinese fabs to 7nm and older process nodes for the foreseeable future.
The US Department of Justice disrupted over $160 million worth of AI chip diversions to China through Operation Gatekeeper in December 2025, and arrested a Super Micro co-founder in March 2026 for alleged chip smuggling, according to Bloomsbury Intelligence. These enforcement actions underscore persistent gray-market attempts to circumvent controls, but the scale of legitimate domestic production now dwarfs smuggling channels.
The Strategic Blunder Hypothesis
White House AI Czar David Sacks acknowledged the policy calculus in remarks to TT News: “That was part of our calculation, of selling not the best but lagging chips to China, is that you can take market share away from Huawei, but I think the Chinese government has figured that out.”
NVIDIA CFO Kress warned in the February earnings call that “our competitors in China, bolstered by recent IPOs, are making progress and have the potential to disrupt the structure of the global AI industry over the long term.” Bernstein analysts project NVIDIA’s China AI GPU market share could fall from 66% in 2024 to roughly 8% in coming years—though Huang’s zero-share confirmation suggests even that forecast was optimistic.
- NVIDIA’s permanent market loss validates narrow containment goal but eliminates future commercial leverage over Chinese AI development
- Domestic chip adoption creates self-reinforcing ecosystem: DeepSeek V4 optimisation for Huawei hardware incentivises further model development on Chinese chips
- China’s 70% self-sufficiency target by 2028 appears credible given foundry consolidation and $70B subsidy commitment
- US Export Controls may have triggered exactly the technological independence they aimed to prevent
What to watch
Huawei’s ability to deliver the projected 750,000 Ascend 950PR units in 2026 will test whether China can scale domestic production to meet hyperscaler demand. Supply constraints remain a bottleneck—ByteDance, Alibaba, and Tencent orders are competing for limited wafer capacity at SMIC’s mature-node fabs. Any production shortfalls could revive gray-market imports or create pressure for policy recalibration.
Beijing’s response to the Trump administration’s December 2025 H200 license grants—blocking imports despite US approval—signals China is willing to accept short-term performance gaps to cement long-term supply chain independence. If Huawei’s next-generation Ascend chips close the 60% performance gap to NVIDIA’s leading edge, the window for US leverage may close permanently.
The strategic question is no longer whether US export controls can slow China’s AI development, but whether they have accelerated the timeline to a bifurcated global AI ecosystem with incompatible hardware and software stacks. NVIDIA’s zero market share is not just a revenue loss—it may mark the point where Tech Decoupling became irreversible.