The Consequences of Exporting AI Chips to China

The Consequences of Exporting AI Chips to China

Nvidia CEO Jensen Huang makes a keynote speech at Computex in Taipei, Taiwan, May 19, 2025.
Nvidia CEO Jensen Huang makes a keynote speech at Computex in Taipei, Taiwan, May 19, 2025. Ann Wang/Reuters

Three CFR experts discuss President Donald Trump’s decision to allow Nvidia to sell advanced AI chip sales to China and what implications it could have for the future of AI, U.S. national security policy, and Chinese relations.

December 9, 2025 2:59 pm (EST)

Nvidia CEO Jensen Huang makes a keynote speech at Computex in Taipei, Taiwan, May 19, 2025.
Nvidia CEO Jensen Huang makes a keynote speech at Computex in Taipei, Taiwan, May 19, 2025. Ann Wang/Reuters
Expert Brief
CFR scholars provide expert analysis and commentary on international issues.

President Donald Trump announced on December 8 that the United States would allow Nvidia to sell its powerful H200 chip—a critical piece of technology in the artificial intelligence (AI) race—to China. 

More From Our Experts

The president said in a social media post that the Commerce Department was negotiating the final details, and he said that other tech companies like Intel would be offered a similar opportunity. Trump also noted that the chips would only be sold to approved U.S. customers and that 25 percent of the revenue would go to the U.S. government as a tax.

More on:

United States

Artificial Intelligence (AI)

China

A bipartisan group in Congress has attempted to derail or limit the sale of AI chips to China since it remains the United States’ most powerful strategic competitor. But Nvidia’s founder and CEO Jensen Huang and other tech allies of the president reportedly pressed the White House to address the export restriction. 

Given the reversal of U.S. national security policy that has sought to limit China’s AI efforts, CFR turned to three of its experts on China and tech policy to better understand the move and its potential consequences. 

Nvidia AI Chip Decision Undermines U.S. Geopolitical Competition With China

Michael C. Horowitz is a senior fellow for technology and innovation at the Council on Foreign Relations and director of Perry World House at the University of Pennsylvania.

More From Our Experts

In the wake of a new National Security Strategy that backed off from characterizing China as a strategic U.S. competitor across the board, the Trump administration’s decision to allow Nvidia to sell H200 chips to China further illustrates that the United States is taking a more dovish approach toward what has been its preeminent geopolitical rival.

Building on the China pressure campaign of Trump’s first term, President Joe Biden’s administration had worked to maintain U.S. leadership in advanced technologies like AI, strengthened diplomatic frameworks such as the Quad to encourage peace in the region, and pursued military capabilities designed to deter Chinese aggression against Taiwan and elsewhere. But that playbook is now getting set aside. The chip decision marks a substantial shift in U.S. technology policy toward China and indicates that one side is currently winning the larger internal Trump administration debate over how to view Beijing.

More on:

United States

Artificial Intelligence (AI)

China

It is indisputable that the decision aids a competitor in the short-term. China’s Premier Li Qiang, DeepSeek’s CEO, and other Chinese tech leaders have explicitly pointed to the lack of Nvidia chips as limiting China’s AI development. The decision will not just further China’s ability to build frontier AI models, it will also support China’s efforts to scale data centers that could directly challenge U.S. companies seeking to compete for international markets. So while the export of a single Nvidia chip—and not even their most advanced chip, the Blackwell—could appear minor to some U.S. observers, it is a major development for China.

While AI competition between the United States and China for market share within both countries and around the world will be a marathon, not a sprint, military competition could peak sooner—and AI capabilities will be critical whenever that occurs. Many U.S. military leaders believe that Chinese President Xi Jinping is seeking to build the capacity to invade Taiwan by 2027, though whether Beijing intends to do so remains unclear. Research shows that China’s military is actively seeking Nvidia chips to fuel AI-enabled military capabilities.

Of course, especially since early reporting suggests China may reject this unilateral concession, as they did with the export of Nvidia H200 chips, China could come back demanding the Blackwell or something else. Far from being over, the conversation about AI chip exports to China could return or even intensify in the coming months or years. And to the extent the decision reflects a broader Trump administration shift towards China, a policy shift back could always lead to the reimposition of controls.

Nvidia’s Chips Will Supercharge China’s ‘Factories of the Future’

Chris McGuire is a senior fellow for China and emerging technologies at the Council on Foreign Relations.

The Trump administration’s decision to allow U.S. firms to sell advanced AI chips to China, including the Nvidia H200 chip, represents a significant change in U.S.-China technology policy. Since 2022, U.S. policy has been to maintain “as large of a lead as possible” over China in AI by restricting Beijing’s access to advanced AI chips. Lifting export controls on H200 chips effectively ends this policy and risks turbocharging China’s development of the most advanced AI models and infrastructure—just as leading AI models are beginning to demonstrate significant strategic capabilities. 

Export controls on AI chips are the only U.S. policy capable of slowing China’s AI progress. These chips are the most important input for developing and running advanced AI models, and China’s domestic production is severely constrained. In every other element of the AI stack—data, talent, algorithms, applications, and electricity generation—China matches or surpasses the United States. But the U.S. lead over China in AI hardware is significant and is rapidly expanding. Liang Wenfeng, the chief executive officer of Chinese AI company DeepSeek, as well as China’s Premier Li Qiang have both publicly cited U.S. export controls on AI chips as China’s biggest constraint in AI development. 

China needs U.S. AI chips to remain competitive. The H200 is not Nvidia’s best chip, but it is more than six times more powerful than the best U.S. AI chip available in China today and better than any chip that leading Chinese chip designer Huawei plans to make for at least two years—considered an eternity in AI development. AI chips are uniquely hard for China to produce: according to Huawei’s own public roadmap, the AI chips that it plans to make next year will decrease in performance relative to its best chips today. And because Huawei’s ability to produce large numbers of these inferior chips is also constrained, the total amount of AI computing power that Huawei can produce per year is only a small fraction of what Nvidia can produce per year—and this gap is widening. 

The United States has declined to export previous-generation AI chips to China for two reasons:

  • large numbers of lower-performing AI chips can be aggregated to create world-class AI data centers capable of training the most advanced AI models, and
  • if China has a larger supply of AI chips, it could start building data centers globally that compete with U.S. AI infrastructure—which to date it has not.  

If Nvidia exports three million H200s to China next year—which is possible and would be broadly consistent with the percentage of Nvidia’s revenue it received from China before export controls went into effect—both fears could come to pass. This would at least triple the amount of aggregate AI computing power China could add domestically next year. This could cause DeepSeek and other Chinese AI developers to close the gap with leading U.S. AI labs and enable China to develop an “AI Belt and Road” initiative—a complement to its vast global infrastructure investment network already in place—that competes with U.S. cloud providers around the world. 

It is paradoxical that while the Trump administration is aggressively using tariffs to reshore manufacturing, it is simultaneously lifting export controls that restrict China’s development of AI data centers—which Nvidia CEO Jensen Huang has repeatedly referred to as the “factories of the future.”

China Is Happy to Take Advantage of Trump’s Open Door

Zongyuan Zoe Liu is Maurice R. Greenberg senior fellow for China Studies at the Council on Foreign Relations. 

In the first half of his first term, Trump began to orchestrate a global tech campaign against Chinese companies Huawei and ZTE. In 2017, the U.S. Congress began limiting the use of Huawei/ZTE gear in critical defense infrastructure. By April 2018, the U.S. had imposed a near-total export ban on U.S.-origin components to ZTE, effectively crippling its capacity. A few months later, the National Defense Authorization Act for Fiscal Year 2019 forbade federal agencies from procuring equipment or services that use Huawei or ZTE components—a major formal statement of distrust.

These measures were meant to slow China’s technological advancement by choking off access to leading-edge U.S. components and severing ties in global supply chains, especially for telecom infrastructure and, by extension, 5G and advanced network technologies.

Fast-forward to Trump’s second term and December 2025, the administration has adopted a dramatic reversal of the export-control posture that they helped build over nearly a decade, effectively undoing what the president started in his first term.

The sale of advanced Nvidia H200 AI chips to approved customers in China does more than signal policy inconsistency: It undermines much of the original purpose of the restrictions. By re-opening the flow of powerful computing hardware to China, Washington risks supplying exactly the tools it once tried to withhold.

But the damage is deeper and more strategic. First, by imposing and then relaxing restrictions, the United States has forced China to accelerate the development of domestic export-control regimes, self-reliance strategies, and an awareness of U.S. economic and systemic vulnerabilities. Meanwhile, China learns that the U.S. will—under pressure or for expedience—back down. That reduces the leverage of future controls.

Second, even if chips like the H200 are sold to China under tighter scrutiny, what Beijing often seeks is not permanent dependence. It buys today to learn today, with the intention to build tomorrow. China’s goal is to be self-sufficient and no longer needs U.S. chips in the long term, and the faster the better from Beijing’s perspective.

Finally, the strategic relationship between the United States and China is no longer one of cautious coexistence built around shared interests, but a bitter, long-term competition defined by asymmetric retaliation—supply-chain decoupling, export controls, and countless bans. The collapse of trust at a strategic level means even a temporary truce will not end the rivalry.

In short, by swinging from tight containment to conditional reopening, the United States has given China not just a respite but a powerful opening. Contemporary policy undermines the very premise of the export-control regime: slowing China’s tech rise. Instead, it has granted Beijing the chance to learn, adapt, and ultimately to compete on nearly equal footing. The strategic contest may now have no clear winner, only a long grind of escalation, adaptation, and rivalry. What Beijing has proven so far is that its core interests and pursuit of self-sufficiency cannot be stopped or denied by the United States. Chinese leaders have a lot of reasons to believe they are not only winning the trade war but also making progress towards a higher degree of strategic autonomy. 

This work represents the views and opinions solely of the authors. The Council on Foreign Relations is an independent, nonpartisan membership organization, think tank, and publisher, and takes no institutional positions on matters of policy.

Creative Commons
Creative Commons: Some rights reserved.
Close
This work is licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) License.
View License Detail
Close

Top Stories on CFR

Democratic Republic of Congo

In shallowly engaging with Kinshasa and Kigali, Washington does little to promote peace and risks insulating leaders from accountability.

United States

Immigrants have long played a critical role in the U.S. economy, filling labor gaps, driving innovation, and exercising consumer spending power. But political debate over their economic contributions has ramped up under the second Trump administration.

Haiti

The UN authorization of a new security mission in Haiti marks an escalation in efforts to curb surging gang violence. Aimed at alleviating a worsening humanitarian crisis, its militarized approach has nevertheless raised concerns about repeating mistakes from previous interventions.