The trade war on technology escalated Wednesday as Donald Trump imposes a 25% tariff on Nvidia AI chips, a move that heavily impacts units destined for China. The new national security order targets high-performance silicon like the Nvidia H200 and AMD MI325X. While the stated goal is to boost U.S. manufacturing, the mechanism of enforcement specifically squeezes foreign buyers.
A key component of this policy is the interaction between the tariff and export controls. The administration recently mandated that advanced chips manufactured in Taiwan but sold to China must first be shipped to the U.S. for security testing. Once these goods cross the U.S. border for their mandatory stopover, they are subject to the 25% import duty, effectively raising the price of AI development for Chinese firms.
Domestically, the impact is softened by broad exemptions. The White House Fact Sheet specifies that chips imported for U.S. datacenters, startups, and public sector applications will not face the tariff. This ensures that American companies driving the AI revolution can continue to access necessary hardware without a tax penalty, preserving U.S. competitiveness.
The administration’s rationale relies on Section 232 of the Trade Expansion Act, citing the danger of relying on foreign nations for 90 percent of America’s chip supply. The order is part of a wider net of tariffs Trump has deployed, including duties on trucks and branded drugs, to force industries to localize production.
Reaction from the industry has been cautious. Stocks for major chip designers like Nvidia and Qualcomm saw minor declines. While legal experts have previously questioned the constitutionality of taxing exports, the administration is framing this as an import duty on goods entering the U.S., even if their final destination lies elsewhere.