A complex new testing requirement has paved the way for new duties as Donald Trump imposes a 25% tariff on Nvidia AI chips. The administration announced Wednesday that chips entering the U.S.—even those just passing through for mandatory testing—will be subject to the levy. This affects high-performance units like the Nvidia H200 and AMD MI325X.
The backdrop is a nine-month Section 232 investigation concluding that the U.S. relies too heavily on foreign chipmakers. To address this “national security risk,” the administration is using tariffs to encourage domestic production. However, to avoid stalling the U.S. tech economy, broad exemptions were granted to U.S. datacenters, startups, and government agencies.
The real teeth of the policy bite into the China trade. The Trump administration recently ordered that chips destined for China must detour from Taiwan to the U.S. for inspection by third-party labs. Because these chips physically enter the U.S., they are hit with the 25% tariff. This creates a significant financial barrier for Chinese access to AI technology.
This maneuver sidesteps potential constitutional issues regarding taxing exports by framing the fee as an import tariff. It also follows through on Trump’s previous threats to curb Beijing’s chip ambitions. While tariffs on “legacy” chips from China have been delayed, this action takes immediate aim at the cutting-edge sector.
Tech giants like Nvidia and AMD saw minor stock fluctuations following the news. The Semiconductor Industry Association and TSMC have not yet commented on the logistical challenges this will create. The order represents a sophisticated use of bureaucracy to exert leverage over the global semiconductor trade.