Why the AI Industry Is Unmoved By Trump’s Tariff Threats

As President Trump has announced varying tariffs over the last month, tech stock prices have dipped, with investors fearing broad impacts on different parts of the tech sector. Shares of NVIDIA, Taiwan Semiconductor Manufacturing Co (TSMC) and AMD have all wobbled, responding in part to news that Trump might implement a 25% tariff on semiconductors shipped to the U.S. 

The possible tariffs are said to be part of Trump’s plan to try to get more semiconductor and AI-related manufacturing in the U.S. But industry insiders say that they’re not yet changing any of their longterm approaches —and are mostly viewing the tariff threats as simply another input factor in a volatile industry whose prices are constantly shifting. 

“Tariffs are more of a blip as opposed to a strong headwind,” says Scott Almassy, semiconductor lead at PwC. Nazar Khan, the COO and CTO of the data center company Terawulf, adds: “No one’s really changing what they’re doing because it’s just too much guesswork.” 

A complex ecosystem

Trump’s semiconductor tariffs could go into effect as early as April 2, he said. 

In the short term, industry players are working to try to dodge those costs as much as possible. Khan says that Terawulf is trying to expedite deliveries so that they arrive before the tariffs do. For deals that will be realized later this year, business partners are negotiating who will bear the extra cost. “Some suppliers have been like, ‘We’re not taking it—you have to pay it,’” he says. “Others are willing to share in the cost because they want us to sign a deal now rather than waiting to see what happens on the tariffs.” 

As economists have pointed out, the cost of tariffs is often passed directly to the end consumer. 

Read More: What Are Tariffs and Why Is Trump In Favor of Them?

“In the first trade war, the evidence was very consistent, and strikingly so, that 100% of the tax was passed forward to the American buyer,” says Mary Lovely, a senior fellow at the Peterson Institute for International Economics. 

She recently wrote a paper predicting that Trump’s tariff proposals in January on Canada, Mexico and China would cost the typical US household over $1,200 a year. The semiconductor tariffs specifically would likely cause anything with a chip—including gaming devices, cars and smart fridges—to increase in price.

But prices for AI products like ChatGPT may not increase in the same linear way, due to several factors. For one, prices inside the AI world have never been static. The Deepseek breakthrough last month reinforced the longstanding trend that the cost of training and maintaining models decreases periodically as technical processes get more efficient. “The question becomes, was the trend going down so much that the 25% is absorbed?” says Khan. 

The demand for AI also remains extremely high, which means that many companies in the ecosystem are operating with comfortable margins and can weather a percentage change in their profit. “Users are so willing to pay right now that it’s probably less impactful for those folks who are procuring them and building out the clusters,” Almassy says. 

Pressure on Taiwan? 

One of the most dominant forces in the chip industry is Taiwan Semiconductor Manufacturing Co (TSMC), which manufactures the chips that power phones and laptops and are central to the output of key American AI players like NVIDIA. 

One of Trump’s main reasons for the semiconductor tariffs, he said, was to try to bring more of their chipmaking to U.S. territory. In February, he accused Taiwan of taking “our chip business away,” adding: “If they don’t bring it back, we’re not going to be very happy.”

Leaders from both parties have supported this mission, wary of America’s outsized dependence on a specialized product manufactured on an island nation that China seems intent on controlling. 

President Biden tried to use a different approach that included incentives: Last year, the Commerce Department pledged up to $6.6 billion for TSMC to expand its Arizona facilities as part of Biden’s CHIPS Act. But TSMC’s progress on that facility has been stymied by repeated delays related to regulatory complexities and a lack of local expertise. 

Recently, most Taiwanese chipmakers have downplayed the potential impact of semiconductor tariffs, including Vanguard, who said that the tariff’s impact would be “trivial” and that they were not considering setting up shop in the U.S. Last month, Taiwan’s economy minister said that “there is an advantage of technological leadership and that cannot be replaced.” 

Almassy agrees. “Especially at the leading edge of AI, which is where Taiwan is, there is no alternative at the moment,” he says. “Taiwan companies will probably be able to pass a lot of that on to their direct customers, who will then have to figure out what they’re going to do with it.” 

Some of those customers are major American AI companies like NVIDIA and OpenAI, who are leading the country’s progress in the so-called AI arms race against China. Some analysts say that the tariffs could actually slightly hamper those companies’ efforts to maintain their lead, especially as AI labs in countries like France or Japan will be able to buy chips tariff-free. “It does probably hamstring American companies a little bit by increasing their costs,” Almassy says. “And if you’re the leader and your resources cost more, you have further to fall than someone not as advanced.” 

A representative for NVIDIA declined to comment. At the company’s earnings call on Wednesday, an NVIDIA executive said that tariffs were “an unknown until we understand further what the U.S.government’s plan is.” 

Other global impacts

It is possible that the tariffs will cause companies to shift their factory locations. But Almassy says that many of these relocations were underway well before the tariff threats, as companies sought to make supply chains more resilient and head off shortages like the ones that unfolded during the pandemic. “There’s a general movement by the industry to diversify where the supply chain sits,” Almassy says.

In a conversation with Goldman Sachs CEO David Solomon in September, Nvidia CEO Jensen Huang said that if the company’s access to Taiwanese components was disrupted, “we should be able to pick up and fab it somewhere else.” But he added that Nvidia probably wouldn’t be able to achieve the same level of “outperformance or cost.”  

Trump claimed one victory in this regard when Apple announced that it would open an AI-related factory in Texas and spend $500 billion to hire 20,000 people in the U.S. But analysts point out that Apple made similar promises at the beginning of both Biden and Trump’s earlier presidential terms. 

Plenty, of course, could change over the next six months. Trump said that he eventually planned to raise the tariff rate even higher, which could force companies to adapt. But for now, many companies that are part of the semiconductor supply chain are choosing to sit tight and see how everything unfolds. “The market’s moving very fast, and it takes time to develop and build new factories,” Khan says. “So the tariff in and of itself may not be determinative for someone to decide to move a facility from Malaysia or Mexico to the US. They may not have the capabilities, money, understanding, or expertise.”