Trump Reverses H20 Chip Ban, Allows Nvidia Sales to China for Revenue Share
The technology sector is grappling with President Trump’s unexpected new agreement with Nvidia, a deal that marks a significant shift in US policy toward China. Earlier this week, the President announced he would permit Nvidia to continue selling its H20 chips to China, albeit in exchange for a 15 percent share of the revenues. Addressing the press on Monday, Trump characterized the H20 as “obsolete,” adding, “You know, it’s one of those things, but it still has a market. So we negotiated a little deal.”
This unusual, and some might say legally questionable, arrangement represents a striking reversal for the Trump administration, which had previously banned all H20 sales to China earlier this year. Reports indicate the President’s change of heart followed a meeting with Nvidia CEO Jensen Huang, who has consistently argued that allowing Chinese companies to acquire H20s does not pose a threat to US national security.
While superficially appearing as a straightforward case of a powerful executive influencing policy in his company’s favor, the underlying narrative is far more intricate, revealing a complex saga of strategic thinking and shifting priorities. Nvidia itself introduced the H20 last year as a less powerful alternative after the US government, under the Biden administration, prohibited the sale of its more advanced H800 chip to China. This initial ban was part of a broader, ambitious initiative spearheaded by Biden administration officials who believed it was critical for the United States to prevent China from achieving a lead in advanced artificial intelligence development.
Recent insights from Graham Webster, a researcher at Stanford University, shed light on the origins of the Biden team’s decision to curb China’s access to advanced semiconductors. His extensive research, based on interviews with over ten former US officials and policy experts—some speaking anonymously—reveals that the official legal justifications for the controls, often citing military and human rights concerns, did not fully capture the strategic rationale. According to Webster, a core motivation was the belief among several key officials within Biden’s White House and Commerce Department that AI was nearing a pivotal point, or several such points, that could confer substantial military and economic advantages upon a nation. Some even anticipated the imminent arrival of self-improving systems, or so-called artificial general intelligence, making the risk of China reaching these thresholds first too significant to disregard.
Consequently, in the fall of 2022, the Biden administration unveiled sweeping export controls designed to prevent China from acquiring the most sophisticated chips necessary for training powerful AI systems. These measures also targeted specialized equipment essential for Beijing to modernize its domestic chipmaking industry. This marked the beginning of a multi-year endeavor that, as Webster notes, was poised to “reshape relations between the world’s two largest powers and alter the course of what may be one of the most consequential technologies in generations.”
A notable aspect of this policy’s trajectory is how many individuals involved in crafting Biden’s export control policies have since transitioned into influential roles across the AI, computing, and national security landscapes. For example, Jason Matheny, who previously led the White House’s technology and national security policy, now serves as president and CEO of RAND, a prominent think tank frequently advising government clients. Similarly, Tarun Chhabra, who contributed to the National Security Council, now heads national security policy at Anthropic, a leading AI company.
These career progressions help contextualize the diverse viewpoints currently shaping the debate over Nvidia’s H20 chips. In January, Anthropic CEO Dario Amodei publicly advocated for a ban on H20 sales to China. He argued that “well-enforced export controls are the only thing that can prevent China from getting millions of chips,” and are therefore “the most important determinant” of whether the US and its allies will maintain a leading global position in the future. Interestingly, Amodei’s post cited semiconductor research published by RAND. Top officials from Trump’s first administration, such as former deputy national security adviser Matt Pottinger, have articulated views consistent with Amodei’s.
Conversely, figures like David Sacks, designated as Trump’s AI and crypto czar, have contended that “overly restrictive” export controls risk allowing Chinese companies to expand their global market share. For the time being, Sacks’ argument appears to have swayed the current administration. However, it remains uncertain which of these competing perspectives will ultimately prove most persuasive to President Trump in the long run, leaving the future of US tech exports to China in a state of flux.