Trump's Export Tariffs on AI Chips Spark 'Dangerous Precedent'
Donald Trump’s approach to trade policy, long characterized by aggressive import tariffs, appears to be charting new, controversial territory: the taxation of US exports. This shift was subtly foreshadowed by an unusual exchange at the White House, where Apple CEO Tim Cook presented Trump with a meticulously crafted gift: a glass plaque and a 24-karat gold base, emphasizing its “Made in California” and “Made in Utah” origins. This gesture coincided with Apple’s announcement of a $100 billion investment in US manufacturing, securing its exemption from a new US tariff on imported computer chips. This transactional dynamic, reminiscent of Trump’s real estate dealings, now extends to highly sensitive technologies, sparking widespread concern among trade experts.
In a dramatic reversal of an earlier ban, the White House recently announced a deal allowing Nvidia and its competitor Advanced Micro Devices (AMD) to sell certain artificial intelligence chips to Chinese companies. The catch? Both tech giants must surrender 15% of their revenue from these sales to the US government. This agreement followed a closed-door meeting between Trump, Nvidia CEO Jensen Huang, and Apple CEO Tim Cook. The move has swiftly prompted suggestions that Nvidia effectively purchased an exemption from escalating US-China trade tensions.
However, this novel “revenue-sharing” arrangement, which critics contend is more accurately an export tax, faces significant legal and ethical scrutiny. Martin Chorzempa, a senior fellow at the Peterson Institute for International Economics, warns that such a deal creates a “dangerous precedent,” fostering the “perception that export controls are up for sale.” He argues that if licenses, ostensibly granted on national security grounds, can be bought, it could unleash a wave of lobbying for the export of other sensitive technologies, fundamentally destabilizing international trading relations. The Department of Commerce is reportedly still “ironing out” the legality of the deal, which some experts believe may violate US laws or the Constitution.
At the heart of this policy shift lies the intense technological arms race between the United States and China. Nvidia, a company that recently became the first publicly traded entity to reach a $4 trillion valuation, produces the crucial processing chips essential for developing and running AI. For years, the US government has imposed restrictions on the export of advanced AI chips and manufacturing equipment to China, aiming to slow Beijing’s AI advancements and maintain a strategic advantage.
Despite these restrictions, China has made considerable progress. Chorzempa notes that while the US retains a significant computing advantage, China’s leading AI models are merely months behind their American counterparts. This narrowing gap has forced US policymakers to re-evaluate where to draw the line on technology exports. The specific AI chips Nvidia and AMD are now permitted to sell to China are not considered high-end; they are suitable for “inference” on trained models (using existing AI to make decisions) but lack the power to “train” new AI models from scratch (teaching AI from vast data sets). Trump himself described them as “an old chip that China already possesses… under a different label.”
Yet, even the export of these less powerful chips remains a point of contention. Hardline proponents of a restrictive US-China policy argue that any chip sale could still contribute to China’s AI capabilities, potentially eroding America’s lead. Others counter that restricting such chips would be ineffective or even counterproductive. The Trump administration’s solution—demanding companies pay for export privileges—is seen by many across this policy divide as a precarious compromise. Republican Representative John Moolenaar of Michigan stated that “export controls are a frontline defense in protecting our national security, and we should not set a precedent that incentivizes the government to grant licenses to sell China technology that will enhance AI capabilities.”
This transactional approach aligns with Trump’s well-known “Art of the Deal” philosophy. He openly recounted telling Jensen Huang, “I want 20% if I’m going to approve this for you,” before negotiating down to 15%. US Treasury Secretary Scott Bessent has even hinted at expanding this “model and beta test” to other industries, suggesting a broader shift in how the US government might engage with businesses on trade. Julia Powles, executive director of the Institute for Technology, Law and Policy at UCLA, cautions that this could lead to “other quid pro quo” demands from the government, potentially pressuring tech companies on sensitive issues like privacy and security, treating them as mere transactional entities rather than institutions operating under established rules. The Nvidia deal, therefore, is not merely about chips; it signals a potentially fundamental reshaping of US trade policy, blurring the lines between national security, economic leverage, and direct government revenue generation.