Nvidia CEO Jensen Huang has expressed uncertainty about the effectiveness of US export controls on semiconductors to China, raising questions about whether these restrictions serve as effective national security measures in the ongoing artificial intelligence race. In a candid CNBC interview on Wednesday, Huang acknowledged that while the export controls’ security benefits remain unclear, they have significantly impacted Nvidia’s business in the Chinese market.
The CEO’s comments come in the wake of Chinese AI startup DeepSeek’s release of its R1 reasoning model, which demonstrated that innovative software development can potentially circumvent hardware restrictions. When asked whether DeepSeek’s achievements showed workarounds against US-imposed semiconductor sanctions, Huang offered a philosophical perspective: “I think that ultimately, software finds a way.” He elaborated that developers consistently create exceptional software regardless of the hardware constraints they face, whether targeting supercomputers, personal computers, phones, or game consoles.
The financial impact on Nvidia has been substantial. Before export controls were implemented in 2022 and subsequently tightened in 2023, China represented a much larger portion of Nvidia’s revenue stream. For the fiscal year ending January 2023, China accounted for 21% of Nvidia’s total revenue. By the fiscal year ending January 2025, that figure had dropped to approximately 13% — representing a revenue decline of nearly 40% in relative terms. Huang noted that his company’s “percentage revenues in China before export controls was twice as high as it is now.”
Despite these challenges and increased competition from Chinese companies like Huawei, Nvidia continues to demonstrate remarkable financial performance. The company reported another strong quarter on Wednesday, with total revenue reaching $39.3 billion — representing a 78% year-over-year increase from $22.1 billion in the same period last year.
The chipmaker did experience temporary turbulence following DeepSeek’s January announcement. Nvidia saw a dramatic single-day sell-off that erased nearly $600 billion from its market capitalization, dropping the company’s valuation to approximately $2.4 trillion. However, the company has since fully recovered, currently standing at a $3.22 trillion valuation, demonstrating investor confidence in Nvidia’s long-term position in the AI hardware market despite emerging competitive pressures and geopolitical challenges.
Key Quotes
It’s hard to tell whether export control is effective.
Nvidia CEO Jensen Huang expressed uncertainty about US semiconductor export controls to China, questioning whether these restrictions actually serve their intended national security purpose in the AI race, despite their clear negative impact on Nvidia’s business.
Our percentage revenues in China before export controls was twice as high as it is now.
Huang quantified the business impact of export restrictions, revealing that Nvidia’s China revenue has been cut in half as a percentage of total revenue, dropping from 21% to 13% over two fiscal years.
I think that ultimately, software finds a way. Maybe that’s the easiest way of thinking about it.
The CEO’s philosophical observation about DeepSeek’s success suggests that software innovation can overcome hardware limitations, implying that export controls may not effectively prevent AI advancement in restricted countries.
Whether you’re developing software for a supercomputer, or software for a personal computer, or software for a phone, or software for a game console — you ultimately make that software work on whatever system that you’re targeting and you create great software.
Huang elaborated on his belief that talented developers consistently find ways to optimize software for available hardware, regardless of constraints, highlighting the adaptability of AI development in the face of restrictions.
Our Take
Huang’s remarks represent a rare moment of corporate honesty about the unintended consequences of technology policy. While Nvidia must publicly support US national security objectives, the CEO’s uncertainty about export control effectiveness signals a growing recognition that hardware restrictions alone cannot contain AI innovation. DeepSeek’s achievements with the R1 model serve as proof of concept that algorithmic efficiency and software optimization can rival or exceed brute-force hardware approaches. This development should prompt serious reconsideration of current policy frameworks. The question becomes: are we hampering American companies’ global competitiveness without meaningfully slowing adversarial AI development? Nvidia’s remarkable recovery to a $3.22 trillion valuation despite losing significant China revenue demonstrates the company’s resilience, but it also raises questions about how much further ahead the company could be without these restrictions. The future of AI competition may depend less on who has the most advanced chips and more on who develops the most efficient algorithms and training methodologies.
Why This Matters
This story highlights a critical tension in global AI development between national security concerns and technological innovation. Huang’s candid assessment that export controls’ effectiveness is uncertain challenges the prevailing narrative that hardware restrictions can meaningfully slow China’s AI advancement. DeepSeek’s success with its R1 model using reportedly less advanced chips validates this concern, suggesting that software optimization and algorithmic innovation can compensate for hardware limitations.
The implications extend beyond Nvidia’s business interests. If export controls prove ineffective at slowing adversarial AI development while simultaneously harming American companies’ competitiveness and revenue, policymakers may need to reconsider their approach to technology competition. The 40% relative decline in Nvidia’s China revenue represents billions in lost business that could fund further R&D and innovation.
Moreover, this situation illustrates how the global AI race is increasingly about software ingenuity rather than pure hardware superiority. As Huang suggests, talented engineers will “make that software work on whatever system” they have available. This reality may require Western nations to focus more on maintaining software and algorithmic advantages rather than relying solely on chip export restrictions to maintain technological leadership in artificial intelligence.
Recommended Reading
For those interested in learning more about artificial intelligence, machine learning, and effective AI communication, here are some excellent resources:
Recommended Reading
Related Stories
- Jensen Huang: TSMC Helped Fix Design Flaw with Nvidia’s Blackwell AI Chip
- US to ban exports of car software to China and Russia
- Outlook Uncertain as US Government Pivots to Full AI Regulations
- The Artificial Intelligence Race: Rivalry Bathing the World in Data
- Biden hails $20B investment by computer chip maker in Arizona plant