- Lead. The Trump administration is considering new semiconductor tariffs that would extend to finished goods containing chips—laptops, data center servers and gaming consoles—moving beyond the 25% duty on certain AI chips imposed in January.
- Fact. Average PC prices have already risen more than 20% in the first half of 2026 from rising component costs; the proposed expansion would layer duties onto goods that incorporate chips produced outside the United States.
- Stake. Commerce Secretary Howard Lutnick is advocating for conditional tariff relief tied to domestic manufacturing commitments—which would effectively make investment in US chip production the price of preferential market access.
The January tariffs on certain AI chips left a deliberate carve-out: data centers, R&D facilities, startups and consumer devices were explicitly exempted, a concession designed to let the US AI infrastructure build-out continue at pace. Those exemptions are now under review, according to reporting by CNBC. The administration is weighing whether to extend duties to any finished product that incorporates chips manufactured outside the United States—a category that would cover nearly every laptop, server and gaming console on the market.
What Would Be Covered
The proposal, which remains preliminary and fluid, would reach beyond the semiconductor component itself to the goods that incorporate it. Data center servers, laptops and gaming consoles are all within the scope under discussion. The administration is also reportedly considering country-specific quotas, establishing separate tariff rates and import allowances by country of origin.
The mechanism Lutnick favors is conditional: companies that commit to manufacturing chips in the United States would receive tariff relief. TSMC’s existing US facilities already benefit from this logic. Any eventual tariffs, if imposed, are expected to be phased in over time rather than applied at once—though no rates have been formally set and the timeline remains unsettled.
The AI Infrastructure Complication
The timing is uncomfortable for the US technology industry. The AI infrastructure expansion—representing hundreds of billions in capital commitment from Amazon, Microsoft, Google and others—depends on server components largely produced in Taiwan, South Korea and Japan. Tariffs on the servers themselves would raise the capital cost of that investment at precisely the moment Washington is trying to secure domestic AI leadership over China.
TrendForce noted that memory suppliers face limited near-term impact because global supply is severely constrained regardless of tariff policy, with new capacity not expected until 2028. The picture is less insulated for general-purpose PC hardware and AI server racks. Broadcom’s $60 billion in debt financing for AI infrastructure illustrates the scale of commitments that could become more expensive if server hardware costs rise materially.
Industry and Policy Trade-offs
Technology companies have warned that a broad semiconductor levy could raise the cost of AI infrastructure just as Washington is attempting to lock in domestic leadership. The White House’s counter-argument is that existing chip policy has already secured “hundreds of billions of dollars” in domestic investment pledges—evidence, it argues, that tariff pressure functions as an effective industrial policy instrument.
The plans remain in flux. The direction of travel, however, is legible: the January framework, designed as a first step, is being evaluated as a floor rather than a ceiling. Whether the exemptions that allowed the AI build-out to proceed will survive the next policy revision is a question the technology sector is now actively watching.