Trump says he’s letting Nvidia sell advanced chips to China
In a significant policy reversal, President Donald Trump announced on Monday, December 8, 2025, that his administration would permit California-based semiconductor giant Nvidia to sell its advanced H200 computer chips to "approved customers" in China. This move marks a notable shift in the ongoing technological and economic rivalry between the United States and China, particularly concerning artificial intelligence (AI) hardware. The decision, revealed by President Trump via a Truth Social post, comes with specific conditions, most notably that the United States government will receive a substantial 25% cut from the revenue generated by these chip sales to China.
The H200, a specialized Graphics Processing Unit (GPU) from Nvidia, is a powerhouse for large language models and other generative AI applications, crucial for advanced data centers and high-performance computing. While representing a significant leap in processing power and memory bandwidth compared to previous generations, President Trump clarified that the agreement explicitly excludes Nvidia’s even more advanced Blackwell system and its soon-to-be-released Rubin system. This distinction underscores an apparent attempt to balance economic benefit with continued strategic control over the most cutting-edge AI capabilities.
President Trump further stated that he had personally informed Chinese President Xi Jinping of his decision, and the Chinese leader "responded positively," indicating a potential diplomatic breakthrough in an otherwise tense tech landscape. The sales, according to Trump, will be executed "under conditions that allow for continued strong National Security," suggesting robust mechanisms will be put in place to monitor the end-use of these powerful chips. Details surrounding these national security conditions and the overall implementation of the deal are currently being finalized by the Department of Commerce. Importantly, the President also indicated that "the same approach" would be extended to other U.S.-based chip manufacturers, including industry stalwarts AMD and Intel, potentially opening new avenues for them in the lucrative Chinese market under similar revenue-sharing arrangements.
The announcement was met with immediate praise from Nvidia. A company spokesperson lauded the president’s decision, describing it as one that "strikes a thoughtful balance that is great for America." This sentiment reflects Nvidia’s long-standing desire for greater access to the Chinese market, which remains a critical revenue stream despite previous U.S. export controls. The company, which has become the most valuable publicly traded entity in the U.S. with a market capitalization hovering around $4.5 trillion as of Monday, has seen its fortunes soar on the back of the global AI boom, fueled by the insatiable demand for its high-performance GPUs.
In Beijing, a day after Trump’s announcement, a representative of China’s foreign ministry responded to the news during a scheduled press conference. The official emphasized the essential nature of bilateral cooperation between the world’s two largest economies, reiterating that such collaboration is "mutually beneficial." This diplomatic language signals China’s readiness to engage in a more structured technological trade with the U.S., particularly concerning components vital for its rapidly expanding AI sector.
The policy shift comes against a backdrop of years of escalating U.S. restrictions on advanced chip technology exports to China. Both the Biden and Trump administrations had previously implemented stringent controls, citing concerns that these chips could be diverted for military purposes or contribute to Beijing’s ambition to build an indigenous AI industry that could challenge U.S. geopolitical and economic interests. These export limits enjoyed broad bipartisan support in the U.S. Congress, with lawmakers frequently emphasizing national security and the need to maintain America’s technological edge in AI. The previous restrictions targeted not only specific chips but also the equipment and talent required to manufacture them, aiming to hobble China’s progress across the entire semiconductor value chain.
Nvidia, however, has consistently advocated for greater market access. CEO Jensen Huang has been a vocal critic of the U.S. export rules, arguing that they have, in fact, "backfired" by inadvertently compelling China to accelerate the development of its own homegrown AI chip industry. Huang’s perspective suggests that by denying access to advanced foreign technology, the U.S. has inadvertently strengthened China’s resolve and investment in self-sufficiency, potentially creating a more formidable long-term competitor. Huang has reportedly engaged in multiple discussions with President Trump regarding export controls, including a meeting just last week, underscoring the intensive lobbying efforts by the semiconductor industry.
This latest deal is not an isolated incident but rather fits into a broader pattern of the Trump administration’s transactional approach to the U.S. semiconductor industry. Earlier in August, the administration had reportedly planned to grant export licenses to Nvidia and AMD for certain chips in exchange for a 15% cut of revenue from sales to China. Furthermore, over the summer, the administration indicated that the U.S. government would take a 10% stake in Intel in return for federal grants aimed at boosting domestic semiconductor manufacturing. These previous initiatives highlight a consistent strategy under Trump to leverage government policy for direct financial returns or strategic stakes in key American technology companies, framing such arrangements as beneficial for the U.S. economy and national interests.
The decision to allow H200 sales, while excluding Blackwell and Rubin, illustrates a nuanced strategy. The H200 offers a significant upgrade in memory capacity and bandwidth, making it ideal for processing massive datasets required by advanced AI models. Its capabilities represent a considerable enhancement for China’s AI research and deployment, enabling faster training and inference for sophisticated applications. However, the more powerful Blackwell and Rubin systems, which promise even greater leaps in AI processing and interconnectedness, remain off-limits, preserving a technological lead for the U.S. and its allies in the absolute bleeding edge of AI hardware. This distinction suggests a calculated risk, allowing China access to powerful tools for its AI development while attempting to prevent it from reaching parity with the absolute latest U.S. innovations.
The implications of this policy shift are far-reaching. Economically, the 25% revenue cut represents a direct financial windfall for the U.S. government, potentially amounting to billions of dollars annually given the scale of China’s demand for AI chips. This revenue could be directed towards various domestic initiatives, including further investment in U.S. semiconductor research, manufacturing, or even broader national security programs. For Nvidia, it reopens a vital market segment, potentially boosting its already astronomical revenues and solidifying its dominant position in the global AI chip market, even with the revenue share.
From a national security standpoint, the "approved customers" clause and the 25% revenue cut are presented as mechanisms to mitigate risks. The approval process would theoretically allow the U.S. to vet end-users, preventing sales to entities directly involved in military programs or human rights abuses. The revenue cut could also be framed as compensation for allowing a potentially sensitive technology transfer, effectively monetizing a strategic concession. However, critics may argue that any advanced chip sale to China, regardless of conditions, contributes to its overall technological advancement, which could indirectly benefit its military or intelligence capabilities in the long run. The challenge lies in precisely defining and enforcing "national security" in the context of dual-use technologies that have both civilian and military applications.
For China, this decision offers a crucial lifeline for its AI ambitions. While the country has poured vast resources into developing its indigenous chip industry, it still lags significantly behind the U.S. and its allies in manufacturing advanced GPUs. Access to Nvidia’s H200 chips provides an immediate, high-performance solution that can accelerate its progress in areas like large language models, autonomous systems, and scientific computing. This could potentially reduce the immediate pressure on its domestic chipmakers, allowing them more time to refine their capabilities without the urgency of fulfilling critical, high-end AI demands. However, the continued exclusion of Blackwell and Rubin systems means China will still face a technological gap at the absolute forefront of AI hardware, reinforcing its long-term goal of complete chip self-sufficiency.
Geopolitically, this policy adjustment could signal a new phase in U.S.-China tech relations, moving from outright denial to a more managed and revenue-generating form of technological exchange. It suggests a pragmatic approach from the Trump administration, where economic gain and strategic control are intertwined. This "managed trade" in critical technologies could set a precedent for other sectors, potentially leading to similar agreements for other dual-use technologies where the U.S. seeks to both profit from and control the flow of its innovations.
However, the long-term ramifications remain to be seen. The semiconductor industry is characterized by rapid innovation, and what is considered "advanced" today may be commonplace tomorrow. The constant evolution of chip technology will require continuous reassessment of such policies. While the H200 deal offers a temporary resolution, the underlying tensions regarding technological leadership, economic competition, and national security between the U.S. and China are likely to persist, making this a pivotal, yet potentially fleeting, moment in the ongoing global tech rivalry.
