NVIDIA CEO Jensen Huang has cautioned that the United States’ recent restrictions on advanced AI chip exports may have unintended consequences like potentially undermining America’s technological leadership rather than curbing China's progress. Speaking at Computex and in a subsequent interview, Huang referred to the new “AI diffusion rule” as a strategic misstep, warning of its serious financial and competitive ramifications.
Record-Breaking Write-Downs
The export ban on NVIDIA’s H20 AI accelerator to China has already led to major financial losses:
$5.5 billion in unsellable inventory write-offs
$15 billion in lost future revenue from pre-sale orders
$3 billion in forfeited tax contributions that could have supported further innovation
Huang noted that no company has ever announced a write-down of this scale, underscoring the unprecedented hit to NVIDIA’s balance sheet.
Immediate Revenue Impact
Despite reporting a robust $44.06 billion in revenue for its first fiscal quarter, driven by strong data center demand, NVIDIA is now bracing for an $8 billion revenue shortfall in the near term, directly linked to the China export restrictions.
Strategic Risks: Boosting Domestic Rivals
Huang warned that restricting U.S. chip exports may accelerate China’s development of domestic AI chips and ecosystems. Denied access to NVIDIA's hardware and software, including CUDA and TensorRT, Chinese firms are likely to ramp up self-sufficiency.
“If we do not compete with China, theirs will dominate the world,” Huang said.
The Full-Stack Imperative
Huang emphasized that AI progress depends on a tightly integrated stack including hardware, systems, and software working in tandem. Attempting to control one layer (like silicon) only drives innovation in others, ultimately weakening the intended effect of the export controls.
Regulating chips alone, he argued, is like playing whack-a-mole—new software will circumvent restrictions and inspire new hardware designs.
Advocating for Open Diffusion
Rather than building walls, Huang supports an “open diffusion” strategy: exporting U.S. AI platforms widely to embed American technology across global infrastructure. Through expanded access to NVIDIA’s software ecosystem and cloud-based GPU services, the U.S. can maintain long-term leadership while respecting national security controls.
Evolving Policy Landscape
U.S. regulators are now considering updates to the Export Administration Regulations, possibly introducing a tiered licensing framework that distinguishes between civilian and dual-use technologies.
NVIDIA and other major chipmakers are lobbying for carve-outs for non-sensitive tools. Meanwhile, U.S. allies like Japan and the Netherlands are aligning their export regimes to close potential loopholes.
What’s Next
As AI continues to shape the global tech landscape, the future of these export controls could redefine not only NVIDIA’s trajectory but also America’s position in the world’s technology race.
Key developments to watch:
Adjustments by the Biden administration to refine export regulations
Congressional hearings balancing national security with global competitiveness
NVIDIA’s strategic response—whether through expanded R&D or international partnerships
Huang’s Vision: Engagement Over Isolation
For Jensen Huang, the path forward is clear: global engagement, not isolation. By integrating American AI platforms into the foundation of worldwide infrastructure, he believes the U.S. can preserve its edge in the defining technological competition of the 21st century.