NVIDIA CEO Jensen Huang has issued one of his strongest public critiques yet of U.S. export restrictions on advanced semiconductor technology to China, saying the company went from holding 95% of China’s AI chip market to virtually zero in a matter of months due to escalating trade controls.
“We have essentially gone to zero,” Huang said. “I can’t imagine any policymaker thinking that that’s a good idea.”
The comments came during a recent industry forum where Huang discussed the global AI economy, semiconductor competition, and Washington’s ongoing tech decoupling from Beijing.
The U.S.–China Tech War Hits NVIDIA Hard
For years, NVIDIA dominated the Chinese AI computing market through its A100 and H100 chips used to train cutting-edge AI systems. However, in 2023 and 2024, the U.S. government expanded export rules restricting high-performance chips to China, aiming to prevent Chinese military and surveillance use of advanced AI.
Despite NVIDIA complying with U.S. policy by releasing slower, export-compliant alternatives (H20, L20, L2 chips), the latest round of controls blocked even those. The result:
| Metric | Before Restrictions | After Restrictions |
|---|---|---|
| NVIDIA China market share | ~95% | ~0% (AI training chips) |
| China revenue share | ~20% of total | Falling sharply |
| Competitor gains | Minimal | Chinese firms rise |
China Strikes Back—With Its Own Chips
With NVIDIA now frozen out of China, Beijing has accelerated efforts to build its own AI chip ecosystem. Companies like Huawei, Alibaba Cloud, Tencent, and Biren Technology are moving aggressively to fill the vacuum.
- Huawei Ascend 910B is now pitched as a direct competitor to NVIDIA’s H100
- State-backed funds are pouring billions into AI infrastructure
- Chinese firms are stockpiling legacy NVIDIA chips from grey channels
According to Huang, this rapid shift may permanently shut NVIDIA out of China, the world’s second-largest AI market after the U.S.
“When you push a market to find alternatives, they will,” Huang warned. “Once those alternatives mature, they don’t come back.”
A Dangerous Isolation Strategy?
Huang stopped short of criticizing U.S. national security motives but warned that an absolute tech split could harm American companies more than Chinese rivals.
“Losing access to such a large market damages our ability to reinvest in R&D and stay ahead,” he said. “We should be careful not to isolate U.S. technology from the rest of the world.”
Tech industry analysts agree the crackdown could backfire by:
- Weakening U.S. chipmakers’ revenue
- Accelerating China’s domestic chip independence
- Splitting AI development into U.S.–Europe vs. China ecosystems
- Slowing global AI innovation
How Big Is the China Loss for NVIDIA?
China has historically accounted for 15% to 25% of NVIDIA’s data center revenue. Losing that market entirely could mean tens of billions in future losses—especially as AI adoption explodes across Asia.
NVIDIA has attempted to pivot by:
- Increasing production for U.S. cloud providers like Microsoft and Amazon
- Expanding AI partnerships in the Middle East and India
- Building sovereign AI cloud systems for U.S. allies
Yet the void left in China is enormous—and growing.
Washington’s View: Security First
U.S. officials argue the restrictions are essential to prevent China from using AI for military modernization, cyberwarfare, and authoritarian surveillance.
The Biden and Trump administrations both argue that AI chips are now strategic weapons, comparable to missiles or fighter jets. “We will not allow U.S. technology to fuel Chinese military AI,” U.S. Commerce Secretary Gina Raimondo said recently.
But Huang believes there must be a more balanced approach:
“Collaborative competition is better than full separation. The world needs AI, and it should be built responsibly together.”
What Happens Next?
NVIDIA is unlikely to regain full access to China anytime soon. The U.S. is expected to tighten tech controls even further in 2025, especially as AI becomes a central focus of national security policy.
However, Huang hinted that NVIDIA will not exit China entirely—its automotive AI, gaming GPUs, robotics tools, and software ecosystems may still find legal paths into China under export rules.
The long-term risk? A split global AI system where U.S. AI and Chinese AI evolve independently—with massive geopolitical consequences.
Final Word
Jensen Huang has been cautious with political commentary in the past, but his new message is clear: Washington may be winning a national security battle while losing a global AI war. NVIDIA built much of China’s AI infrastructure—but U.S. policy may now be helping China build its own rival AI economy faster than expected.
“We can compete without cutting ties,” Huang said. “Because when you force people to go it alone—they will.”
