The White House revealed its latest trade strategy on Wednesday as Donald Trump imposes a 25% tariff on Nvidia AI chips. The executive order targets specific high-performance processors, including the Nvidia H200 and AMD MI325X, citing Section 232 of the Trade Expansion Act. The move is predicated on the finding that foreign dependence for chips constitutes a national security risk.
The strategy involves a “light touch” for the domestic market. To avoid stifling the U.S. AI industry, the administration has exempted chips imported for data centers, startups, and government use. This ensures that the entities driving the next wave of technological progress in the U.S. are insulated from the tariff’s direct costs.
The policy’s aggressive nature is directed at re-engineering global trade flows. A key provision mandates that chips destined for China must detour through the U.S. for testing, thereby triggering the 25% tariff. This mechanism allows the U.S. to extract revenue from the China trade and impose costs on a strategic rival without formally taxing exports.
At its core, the order is an industrial policy aimed at reshoring manufacturing. With the U.S. producing only 10% of its chips, the administration believes tariffs are necessary to incentivize domestic production. The order acts as a signal to the industry that the days of relying solely on offshore manufacturing are numbered.
The tech industry is processing the implications. While the exemptions offer short-term relief, the administration has made it clear that broader tariffs are possible. The Commerce Secretary’s discretion to modify the rules introduces a layer of uncertainty, forcing companies to seriously consider their manufacturing footprints.