In a significant move for the global tech industry, former President Donald Trump has imposed a substantial 25% import tariff on advanced artificial intelligence (AI) semiconductors, directly targeting flagship products from industry giants Nvidia and AMD. The action, justified on national security grounds, forms part of a broader strategy to reduce American reliance on foreign chip manufacturing.
The National Security Proclamation and Its Targets
The White House released a proclamation on Wednesday detailing the new duties, which follow a nine-month investigation under Section 232 of the Trade Expansion Act of 1962. The tariffs specifically apply to high-performance AI chips meeting certain technical benchmarks, including Nvidia's H200 AI processor and AMD's MI325X accelerator. The order also covers devices containing these targeted semiconductors.
The administration's rationale hinges on a stark statistic: the United States currently manufactures only about 10% of the chips it requires. This heavy dependence on overseas supply chains, particularly from geopolitically sensitive regions like Taiwan, was labelled a "significant economic and national security risk" in the official proclamation.
Broad Exemptions and Industry Impact
Crucially, the tariffs are designed to be narrowly focused. According to a White House fact sheet, they will not apply to chips imported for use in US data centres, which are massive consumers of AI hardware. Further exemptions cover startups, consumer applications not involving data centres, non-datacentre industrial uses, and public sector projects.
US Commerce Secretary Howard Lutnick has been granted broad discretionary power to apply additional exemptions as needed. The immediate market reaction saw shares in Nvidia, AMD, and Qualcomm trade slightly lower in after-hours trading following the announcement.
A Pattern of Trade Actions and Future Implications
This latest action fits into a wider pattern of trade measures aimed at reshoring critical manufacturing. In December, Trump stated he would impose tariffs on Chinese semiconductor imports, delaying that move until June 2027. That decision followed a separate "Section 301" investigation into China's exports of older-technology, or "legacy," chips initiated under the Biden administration.
Notably, the administration has also instituted a new rule requiring China-bound chips made in Taiwan, like Nvidia's H200, to be routed through the United States for third-party testing. Upon entry to the US for this purpose, the chips become subject to the new 25% levy. An annex to the order confirms this tariff will not be stacked on top of other existing Section 232 duties on materials like steel and aluminium.
While offering a reprieve for data centres for now, the White House fact sheet indicates that broader tariffs on semiconductors and derivative products could be imposed in the near future to further incentivise domestic manufacturing. Leading US firms like Nvidia, AMD, and Intel design the world's most sought-after chips, but the vast majority are fabricated overseas, often by Taiwan Semiconductor Manufacturing Co (TSMC).
AMD stated it "complies with all U.S. export control laws and policies." Nvidia and TSMC did not immediately respond to requests for comment on the new tariffs.
