Invoking the powers of the Trade Expansion Act of 1962, Donald Trump imposes a 25% tariff on Nvidia AI chips, marking a significant escalation in tech trade policy. The Wednesday proclamation follows a Section 232 investigation that deemed the current level of semiconductor imports a national security threat. High-end chips like the Nvidia H200 and AMD MI325X are the primary targets.
The investigation highlighted that the U.S. only produces a fraction—approximately 10 percent—of the semiconductors it consumes. The administration asserts that this dependency on foreign nations, particularly Taiwan, creates unacceptable risks. The tariff aims to rectify this by penalizing imports and encouraging domestic fabrication.
However, the administration is mindful of the costs to U.S. innovation. A fact sheet accompanying the order detailed broad exemptions for U.S. datacenters, the public sector, and startups. These entities will not have to pay the 25% duty, allowing the U.S. AI sector to continue its rapid growth unimpeded by new taxes.
The tariffs serve a dual purpose by also targeting China. Chips bound for Beijing must now detour through the U.S. for testing, a process that legally constitutes importation and triggers the tariff. This creates a financial hurdle for Chinese companies attempting to acquire advanced AI technology, complementing other restrictions on tech exports.
This move is part of a larger tapestry of protectionist policies, including recent tariffs on pharmaceuticals and vehicles. While the stock market reacted with a slight dip for chipmakers, the long-term consequences for the global supply chain remain to be seen. The Commerce Department holds the keys to further exemptions, ensuring the policy can evolve.