Nvidia’s CEO, Jensen Huang, has voiced strong reservations regarding the efficacy of the U.S. government’s restrictions on exporting advanced artificial intelligence (AI) chips to China. He contends that these measures, initiated under the Biden administration, have inadvertently stimulated the growth of China’s domestic AI industry while simultaneously impacting Nvidia’s financial performance.
The Policy’s Unintended Consequences
Huang elaborated on his perspective during an event in Taiwan, asserting that the export controls, intended to slow China’s AI advancements, have instead backfired. He pointed to a significant shift in Nvidia’s market share within China, remarking, “About four years ago…Nvidia’s market share in China was nearly 95%. Today, it’s only 50%. The rest was built by Chinese technology.” This suggests that the restrictions have spurred Chinese companies to develop their own AI chip technologies, reducing their reliance on U.S. suppliers.
Moreover, Huang highlighted the economic repercussions for Nvidia, citing the need to sell less advanced chips to comply with the export regulations. This shift has led to lower average selling prices and, consequently, reduced tax revenues for the U.S. government. He emphasized that the policy has failed to impede China’s AI research efforts. Instead, Chinese researchers have innovated and developed their own technological solutions, mitigating the impact of the export controls. The situation highlights a critical issue in technology policy: the potential for unintended consequences when attempting to restrict access to key technologies. The assumption that limiting access would stifle innovation has proven to be inaccurate in this case, as it instead galvanized domestic development within China. This suggests a need for more nuanced and adaptive approaches to technology regulation, focusing on fostering innovation and competitiveness within the U.S. while addressing legitimate security concerns.
The consequences extend beyond mere market share losses for Nvidia. The reduced revenue streams also impact the company’s ability to invest in future research and development, potentially hindering its long-term competitive advantage in the global AI market. This ripple effect underscores the interconnectedness of technology policy, corporate strategy, and national competitiveness. Furthermore, the development of indigenous AI chip technologies in China could accelerate the country’s broader technological autonomy, reducing its dependence on foreign suppliers across various sectors. This strategic shift could have significant geopolitical implications, altering the balance of power in the global technology landscape.
The need for U.S. policymakers to re-evaluate their approach to AI chip exports to China is now more critical than ever. A more effective strategy might involve promoting collaboration and knowledge sharing in certain areas of AI research, while focusing on targeted restrictions that address specific national security concerns without hindering the overall progress of the industry. This could involve fostering stronger partnerships with allied nations to develop alternative supply chains and promote open-source AI technologies that are less susceptible to geopolitical manipulation. Ultimately, the goal should be to maintain U.S. leadership in AI innovation while ensuring a level playing field and preventing the unintended consequences of restrictive trade policies.
China’s Technological Rise Amidst Restrictions
Despite the limitations imposed by the U.S., Huang acknowledged the resilience and ingenuity of Chinese companies. He emphasized that Chinese AI researchers are continuing their work and unlocking their native technology. If denied access to Nvidia’s products, they are finding alternatives, either through their own innovations or by utilizing “the second best” options available, indicating a level of self-sufficiency that undermines the desired effect of the export restrictions. This points to the adaptability of Chinese technology companies in the face of adversity. Their ability to quickly develop and deploy alternative solutions highlights the depth of their technical expertise and the strength of their domestic ecosystem.
Furthermore, the availability of “second best” options suggests that the U.S. export controls are not preventing Chinese companies from accessing the necessary computing power to pursue their AI research goals. Instead, they are simply incentivizing them to rely on domestically produced or alternative foreign-sourced technologies. This undermines the intended effect of slowing down China’s AI advancement and may even accelerate its progress by forcing it to become more self-reliant.
The US restrictions might even inadvertently be pushing Chinese companies away from established western standards and practices in AI technology. This could lead to the development of a parallel AI ecosystem with its own unique standards, architectures, and even ethical frameworks. This divergence could create compatibility issues and hinder future collaboration between Chinese and Western AI communities, further isolating China and potentially leading to a fractured global AI landscape.
The situation underscores the importance of understanding the dynamics of technological innovation and the limitations of purely restrictive policies. While export controls may be useful in certain limited circumstances, they are often ineffective in preventing the diffusion of knowledge and the development of alternative solutions. A more comprehensive approach to technology policy should focus on fostering innovation, building competitive advantages, and promoting collaboration, rather than simply attempting to block access to specific technologies.
Recognition of Chinese AI Leaders
The CEO also took the opportunity to commend specific Chinese companies, notably Hangzhou DeepSeek Artificial Intelligence Basic Technology Research Co Ltd (DeepSeek) and Huawei Technologies Co Ltd, for their significant contributions to the AI landscape. He recognized DeepSeek’s work, emphasizing its open-source nature and its functionality to Nvidia’s technology. Huang’s recognition of DeepSeek and Huawei represents a notable shift in the narrative surrounding Chinese technological innovation. By acknowledging their contributions to the AI landscape, he implicitly challenges the perception that Chinese companies are simply imitators or copycats. Instead, he presents them as innovators and key players in the global AI ecosystem.
Specifically, Huang underscored DeepSeek’s “reasoning model,” which allows machines to think, reason, plan, and read, describing it as “incredible for AI infrastructure.” Open-sourcing this technology has allowed developers around the world to use and build upon DeepSeek’s advances, exponentially increasing computational needs. This highlights the power of open-source collaboration in accelerating innovation. By making their technology freely available, DeepSeek has enabled a global community of developers to contribute to its improvement and expand its applications. This open-source approach also benefits Nvidia by driving demand for its high-performance computing infrastructure, as developers require powerful hardware to run and experiment with DeepSeek’s models. The rise of open-source AI models like DeepSeek’s could democratize access to advanced AI capabilities, enabling smaller companies and researchers to participate in the AI revolution without having to invest massive resources in building their own proprietary models. This could foster greater innovation and competition in the AI market, leading to new and unexpected breakthroughs.
Huang acknowledged Huawei as one of the world’s “largest and most formidable” technology companies. He praised their rapid innovation and highlighted the unique advantages of AI infrastructure, where multiple chips can be combined to overcome limitations, unlike in devices like cellphones. Huawei’s resilience in the face of U.S. sanctions is a testament to its technological capabilities and its ability to adapt to challenging circumstances. The company’s continued innovation in areas such as AI, 5G, and cloud computing demonstrates its commitment to remaining a leading player in the global technology market.
The advantages of AI infrastructure, where multiple chips can be combined to overcome limitations, are particularly relevant in the context of export controls. By focusing on building large-scale AI infrastructure, Chinese companies can overcome the limitations imposed by restrictions on individual chips. This highlights the strategic importance of investing in infrastructure that can leverage the power of distributed computing to achieve performance levels comparable to those achieved with more advanced individual chips.
The AI Infrastructure Advantage
Huang believes that China’s abundance of energy and land resources positions it favorably to develop AI infrastructure using less-advanced chips. Consequently, he does not believe the US ban on Nvidia’s H20 chips in China is effective. China’s advantages in energy and land resources provide it with a unique opportunity to build cost-effective AI infrastructure. The availability of cheap energy allows Chinese companies to operate large data centers at a lower cost than their counterparts in other countries. Similarly, the abundance of land provides ample space for building these data centers, which require significant physical footprint.
He predicted that “The Chinese would just buy more chips from Huawei and others, because we are not the only ones who could provide such technology. Many others are more than happy to do that.” This competition suggests that the export controls are simply diverting business to other players, without fundamentally hindering China’s AI development. Huang’s prediction highlights the futility of attempting to stifle China’s AI development by restricting access to Nvidia’s chips. The existence of alternative suppliers, both domestic and foreign, ensures that Chinese companies will be able to obtain the necessary computing power to pursue their AI goals. This also points to a potential for increased competition in the global AI chip market, as other players seek to capitalize on the opportunities created by U.S. export controls.
The situation raises questions about the effectiveness of unilateral export controls. While the U.S. may be able to restrict access to its own technologies, it cannot control the availability of technologies from other countries. This highlights the importance of international cooperation in addressing concerns related to technology proliferation. A multilateral approach, involving collaboration with allied nations, would be more effective in preventing the unauthorized transfer of sensitive technologies and ensuring a level playing field in the global technology market.
Huang expressed hope that Nvidia could regain its market share in China, pending a shift in policy: “The US government now realises that. So, we hope that we can go back to China to win the market share as soon as possible.” Huang’s hope reflects Nvidia’s strong interest in maintaining its presence in the Chinese market. China is a key market for Nvidia, and the company is eager to regain the market share it has lost due to export controls. This underscores the economic importance of the Chinese market to U.S. technology companies and the potential costs of restrictive trade policies.
Trump’s Shift in Perspective
Huang commended former US President Donald Trump’s decision to reverse the AI Diffusion Rule, which was initially implemented by the Biden administration. Huang pointed out Trump’s realization that the US is not the only source of AI technology. Trump’s decision to reverse the AI Diffusion Rule signaled a recognition of the limitations of unilateral export controls and the importance of promoting innovation and collaboration in the AI field. By acknowledging that the U.S. is not the only source of AI technology, Trump implicitly recognized the need to compete on a level playing field and to avoid policies that could stifle innovation and harm the U.S.’s competitive advantage.
He elaborated, “The goal of the old AI Diffusion Rule in the past was to limit AI diffusion, and President Trump realised this was a wrong goal, because the US is not the only provider of AI technology,” arguing that restricting the spread of AI knowledge and tools would ultimately harm the U.S.’s competitive edge. Huang’s argument highlights the importance of maximizing the diffusion of AI technology to foster innovation and economic growth. Restricting the spread of AI knowledge and tools would only serve to hinder the development of AI in the U.S. and to give other countries a competitive advantage. This underscores the need for a policy approach that promotes innovation, collaboration, and the free flow of information, while addressing legitimate national security concerns in a targeted and effective manner.
The premise that attempting to completely contain cutting-edge technological innovation within national borders is not only unrealistic but also damaging to the innovating nation’s overall progress is increasingly self-evident in the modern interconnected world. A far more fruitful approach lies in embracing open standards, nurturing international collaborations on non-sensitive research, and aggressively securing and defending intellectual property through robust legal frameworks.
Maximizing AI Diffusion
Huang believes that the U.S. must maximize AI diffusion to maintain its leadership in AI advancements and encourage global adoption of American technology. Maximizing AI diffusion is essential for maintaining U.S. leadership in AI advancements. By promoting the widespread adoption of U.S. AI technology, the U.S. can create a virtuous cycle of innovation and economic growth. This requires a policy environment that encourages investment in AI research and development, promotes collaboration between industry and academia, and fosters a skilled workforce capable of developing and deploying AI solutions.
Huang stated, “That’s where we are today. It’s really a great reversal of a wrong policy, and it’s just in time, but we need to move fast now,” emphasizing the need for swift action to capitalize on the changed policy landscape. Huang’s statement underscores the urgency of taking action to capitalize on the changed policy landscape. The U.S. must move quickly to implement policies that promote AI diffusion and ensure that it remains a leader in the global AI race. This requires a concerted effort from government, industry, and academia to invest in AI research and development, foster a skilled workforce, and create a regulatory environment that encourages innovation.
Huang reiterated that the assumption underlying Biden’s AI Diffusion Rule was “completely proven fundamentally flawed.” The flawed basis of the policy inevitably led to flawed outcomes, necessitating a change in direction. The failure of the AI Diffusion Rule underscores the importance of basing policy decisions on sound evidence and a thorough understanding of the dynamics of technological innovation. Policies that are based on flawed assumptions are likely to lead to unintended consequences and to undermine the goals they are intended to achieve. This highlights the need for a more evidence-based approach to technology policy, one that is informed by data, analysis, and a deep understanding of the complex interplay between technology, economics, and national security.
Huang added, “That’s why Trump made it possible for us to extend our reach outside the US. He said publicly that he would like Nvidia to sell as many GPUs (graphics processing units) as possible around the world,” underlining the potential for Nvidia to contribute to global AI development. Huang’s statement highlights the potential for U.S. companies like Nvidia to contribute to global AI development. By selling their products and services around the world, these companies can help to spread AI technology and to foster innovation in other countries. This can lead to a more collaborative and interconnected global AI ecosystem, which benefits everyone.
China’s Strategic Importance
Huang consistently emphasized China’s critical role in Nvidia’s overall growth strategy. The Chinese market contributed US$17 billion in revenue, representing 13% of Nvidia’s total sales, in the financial year ending January 26, 2025. China’s strategic importance to Nvidia’s growth strategy is undeniable. The Chinese market is a major source of revenue for Nvidia, and the company is committed to serving its customers in China. This underscores the economic importance of the Chinese market to U.S. technology companies and the potential costs of restrictive trade policies.
The US government has restricted exports of Nvidia’s most advanced chips to China since 2022, prompting the development of modified chips like the Hopper H20. However, Huang indicated that future Nvidia chips for China would not be modifications of the Hopper series, suggesting an exploration of new architectures and technologies to serve the Chinese market. The development of modified chips like the Hopper H20 demonstrates Nvidia’s commitment to complying with U.S. export controls while continuing to serve its customers in China. However, Huang’s indication that future Nvidia chips for China would not be modifications of the Hopper series suggests that the company is exploring new architectures and technologies to better serve the Chinese market. This could involve developing chips that are specifically designed to meet the needs of Chinese customers, while also complying with U.S. export controls.
However, Huang reportedly said that after the H20, the next Nvidia chip for China will not be from the Hopper series, because “it’s not possible to modify Hopper anymore.” This statement suggests that the limitations of modifying existing architectures to comply with export controls are becoming increasingly restrictive. This could lead Nvidia to develop entirely new chip architectures specifically designed for the Chinese market, potentially leading to further divergence in chip technology between the U.S. and China.