AI industry critical of Trump's plan to impose heavy tariffs on chips

The Trump administration is considering new tariffs on semiconductors and a wide range of end products containing chips, with potential rates of up to 100 percent. Politico reported this on 27 August 2026. The tech industry is responding sharply: an unnamed official from a major trade association, who was also active during Trump's first term, called the proposal "the dumbest conceivable way to pursue American dominance in AI."

The plans are striking because they run counter to the policy Trump himself is pursuing to accelerate the expansion of American data centers. In January he signed Executive Order 14318, aimed at reducing regulatory burdens for large data center projects. Now, cost increases threaten to restrain those very investments.

The Computer & Communications Industry Association (CCIA) estimates that such tariffs would reduce US GDP by approximately $90 billion per year and that around 20 percent of planned data center projects through 2030 could be delayed or cancelled.

What the tariffs would specifically cover

In January, Trump already signed a narrower 25 percent tariff on a select group of advanced computer chips, including Nvidia's H200 and AMD's MI325X. Chips imported for use in data centers, research, consumer products, industry, or government were excluded from that levy. Those exemptions are now under review.

The new plans go further. In addition to semiconductors themselves, they could cover derivative products such as data center servers, laptops, gaming consoles, and smartphones. Used or refurbished products may also be subject to the levy. Specific tariff rates have not yet been set, but levels of up to 100 percent are being discussed.

Commerce Secretary Howard Lutnick is said to be advocating internally for an alternative framework: companies would be allowed to import a certain volume of chips tariff-free, on the condition that they commit to manufacturing investments in the United States. Details about exactly how that mechanism would work have not been made public.

Why the industry sees this as a contradictory signal

The core of the criticism is that the US simply cannot currently produce enough advanced chips domestically. More than 90 percent of the world's most advanced semiconductors are manufactured in Taiwan. South Korea and Malaysia also supply a significant share. New factories in the US take years, sometimes decades, and billions of dollars before they reach full capacity.

Jonathan McHale, vice president of digital trade at the CCIA, compared the current rollout of data centers to the construction of the transcontinental railroad in the nineteenth century: an infrastructure project of national strategic importance. Additional costs and tariff uncertainty put pressure on those investments, he argued.

The industry also points out that double taxation looms if both the chip itself and the end product containing it are subject to a tariff. Trade associations have proposed as an alternative that AI server chips be fully exempted and that any tariff of 25 percent be reduced to 10 percent.

Consequences for consumers and businesses

If the plans go ahead, the sector anticipates price increases for a range of devices: smartphones, laptops, tablets, smartwatches, and connected vehicles. The launch of new products with AI features could be delayed if manufacturers face higher procurement costs for components.

For companies building or expanding data centers, the uncertainty itself is already a problem. Investment decisions on large infrastructure projects are made on the basis of long-term costs. Higher or uncertain hardware costs could lead companies to build capacity outside the US, according to warnings from the sector.

The CCIA calculated in June that the GDP impact would be around $90 billion per year. That figure covers the combined effects of higher production costs, delayed investments, and the relocation of projects abroad. The organization published that calculation ahead of the most recent tariff plans; the ultimate scale will depend on which products and tariff levels are established.

Tension within its own policy

Trump has publicly spoken out strongly in favor of rapid data center expansion. He explicitly criticized communities that refuse data centers and signed an order earlier this year to streamline permitting procedures for projects above 100 megawatts. That stance is in conflict with measures that raise hardware costs for precisely those data centers.

The semiconductor industry expects its dependence on imported chips to persist for years, even if new American factories are built on schedule. TSMC is constructing factories in Arizona, but production capacity for the most advanced chips there remains limited. Intel and Samsung are also expanding in the US, but those trajectories also span multiple years.

For the European and Dutch AI scene, this debate is relevant for a different reason. If American data center companies adjust their investment plans due to tariff risks, some of the capacity expansion could end up outside the US. Europe has long been trying to build its own AI infrastructure, and uncertainty in the US could make that option more attractive for some companies. At the same time, European chipmakers and data centers are also dependent on the same global supply chain, and tariff escalation in the US will feed through into global component prices.