Trump semiconductor tariffs could broaden to data-center servers
A reported Trump tariff plan could cover chips and some products containing them, raising fresh concerns for AI data-center builders.
By Renata Fuchs · Policy Reporter
· 3 min read
Proposed Trump semiconductor tariffs could reach chips and selected downstream products, including laptops, gaming consoles and data-center servers, according to Politico’s reporting from people familiar with the private discussions. The framework is not final, no tariff rate has been settled, and sources said it could change in the coming weeks or months.
For companies building AI infrastructure, the Trump semiconductor tariffs data centers question is whether the administration will exempt servers and the imported chips needed to add computing capacity. A broader tariff base could raise the cost of imported servers or imported chips used in new data-center capacity while demand for advanced processors is already high.
Politico reported that Commerce Secretary Howard Lutnick favors linking tariff relief for foreign companies to investment in US chip manufacturing. Four people familiar with the discussions said the model under consideration would allow a defined volume of chips to enter duty-free, with that allowance tied to companies’ commitments to manufacture in the United States. Officials have also considered phasing in the duties, Politico reported.
What could Trump’s semiconductor tariffs cover?
The possible scope goes beyond standalone semiconductors. People familiar with the deliberations told Politico that laptops, gaming consoles and servers that populate data centers could be covered. The precise products, rates and country-specific treatment remain unresolved, according to the report.
The White House defended its trade-policy approach. Spokesperson Kush Desai said reshoring semiconductor production is a priority for President Donald Trump and said administration policies have secured hundreds of billions of dollars of investment in the sector. That investment figure is the White House’s claim. Politico said the Commerce Department did not respond to multiple requests for comment.
Why are technology companies seeking data-center exemptions?
Industry representatives argue that the timing conflicts with the administration’s stated aim of US leadership in AI. Advanced chip fabs cost billions of dollars and take years, or potentially decades, to build, sector advocates told Politico. Until more domestic capacity comes online, US companies remain dependent on overseas chip suppliers.
Technology companies have been lobbying the administration for exemptions, particularly for data centers. But three people familiar with the talks told Politico that discussions have recently become less favorable to the industry, with officials signaling that exemptions included in January tariffs may no longer be supported for data centers, research and development, startups, consumer applications and other uses.
Jonathan McHale, digital policy chief at the Computer and Communications Industry Association, said greater costs and less predictable conditions would make data-center investment harder. CCIA includes Amazon, Google and Meta among its members.
One unnamed official at a major industry group, who also served in Trump’s first administration, called the reported plan “the single dumbest way imaginable” to pursue US AI leadership. That is the view of one unidentified official, rather than a stated position shared by the entire AI industry.
CCIA has separately estimated that applying semiconductor tariffs to data centers could reduce US GDP by about $90 billion annually and delay or cancel roughly 20% of data-center projects planned through 2030. Those figures are the trade association’s projections, not established outcomes. The open question for operators is whether a final policy preserves exemptions during the current buildout or makes imported compute infrastructure more expensive before domestic supply can replace it.
This story draws on original reporting from Ars Technica.