President Donald Trump promised to make America the “world capital” of artificial intelligence, but his belligerent trade and tariff actions are set to undercut Silicon Valley’s efforts on the vital technology and its rivalry with China.
Trump’s tariffs on trade will increase the expense of building, outfitting, and running the data centers that companies like OpenAI, Google and Microsoft are competing to build to fuel AI development, executives and AI and data center construction experts say.
Late Friday, Trump suddenly exempted computers, smartphones and the high-powered semiconductors called GPUs that are essential to AI projects from his tariffs. He had already indicated there would be no exceptions.
The action seemed at first to bring great relief to the tech sector, but numerous ingredients for AI data centers like building materials, cooling systems and generator backup are still subject to import duties. And the exemptions don’t seem to counteract the 20 percent tariff Trump imposed on Chinese imports this spring.
Additionally complicating matters, Commerce Secretary Howard Lutnick said on Sunday that tariffs on semiconductors are still being developed and will be applied through Section 232, a provision covering tariffs dealing with national security and calling for a lengthy process of study and comment.
OpenAI CEO Sam Altman, in an interview before Trump’s new exemptions were announced, said the company was urgently trying to figure out how the tariffs would affect the cost of running its AI models. “We’re working around-the-clock on this,” he said.
Trump and tech industry leaders view China as America’s main competitor in AI, and the president has indicated that the U.S. must remain ahead to protect its economic and national security. Upon assumption of office, he moved swiftly to roll back AI regulations enacted under the Biden administration and simplify the construction and powering of AI data centers.
But Trump’s draconian Chinese tariffs — currently eye-watering at 145 percent — can even help it outcompete America in AI. The nation is a primary supplier of data centre equipment, such as the equipment not just newly exempt from some of Trump’s tariffs.
“As the administration continues to weigh tariff policies, we strongly urge efforts to bring certainty and ongoing consideration of impacts on key data center equipment and components at this critical juncture in the AI race,” Josh Levi, president of the Data Center Coalition, a trade association representing companies that own data centers, said in a statement issued prior to the exemptions being announced.
The data center building and AI industry are central to America’s national security and global competitiveness, Levi stated.
“It would sound good to have chips exempt, but then there are so many other components to a data center,” Altman said in an interview Thursday after delivering remarks at the Vanderbilt Summit on Modern Conflict and Emerging Threats.
The potentially steep cost escalation resulting from Trump’s tariffs, along with the unsettled and rapidly shifting trade policy, has created a wave of uncertainty in an industry that tends to view itself as boldly shaping the future.
The broader observation is that of uncertainty and confusion and lack of ability to actually plan,” said Jay Biggins, executive managing director at BLS and Co., a real estate consulting firm that assists AI data center developers with site selection and supply chain planning. He was speaking before the exemptions were announced on Friday.
For decades, the tech industry has been captivated by artificial intelligence, but the introduction of ChatGPT by OpenAI in 2022 ignited a fierce competition to advance AI technology and explore profitable avenues for delivering it to both consumers and enterprises.
The AI boom hasn’t yet produced massive profits, but Big Tech companies and venture capitalists have invested billions of dollars to build and operate AI models. They’ve created a data center building frenzy throughout the United States. The gargantuan power use of the facilities has strained power utilities and prompted efforts to restart idled coal and nuclear facilities.
Google has indicated that it will invest $75 billion this year in AI data centers, while Microsoft is targeting $80 billion. On the first complete day of his second term, Trump held a public announcement in the Oval Office featuring OpenAI’s Altman alongside executives from Oracle, a leading business software firm, and Japan’s SoftBank, who pledged to invest up to $500 billion in AI data center development as part of a project named Stargate during his term.
“The AI data center market is probably the largest market I’ve ever seen in my career” when it comes to construction, said Don Clark, co-CEO of Clark Pacific, a Sacramento-based prefabricated building company that has a growing line of business constructing data centers.
Several within the industry fret that Trump’s tough tariffs strategy may limit that boom.
Firms at the forefront of the AI boom saw their share prices trashed by Trump’s initial tariff declaration but have mostly regained ground in recent days.
Representatives at Microsoft, Nvidia, and Google did not comment. The high demand for data center equipment and extensive order backlogs would make it difficult for American companies to redo their strategies for the Trump tariff era.
Even if corporations are able to obtain generators, switches, and transformers from domestic suppliers, the demand is already so great that prices will need to go up, according to Biggins, the real estate advisor.
“All of the necessary hardware was already on 24-, 36-month backlogs,” Biggins stated. Data center prices might increase by 15 percent or 20 percent, he added. Certain U.S. AI firms would rather construct data centers elsewhere in the world, Biggins stated. Ng, the previous Google AI laboratory leader, who has called for the U.S. to invest in AI and not have regulations that could restrain the technology’s growth, concurred.
Tech firms tend to prefer constructing data centers near their clients so that they can retrieve apps, data, and emails with less lag, Ng added. That limitation is not as acute, however, for AI applications, which take more time to process before answering questions from users.
When an AI service such as ChatGPT takes 10 seconds or more to produce an AI image, it’s not a big deal to add on the few milliseconds to transmit it from a remote data center, he said. Big Tech companies already have data centers all around the globe and are putting new ones in places such as Malaysia and Singapore. “When the rules change overnight by tweet, it’s hard to plan,” Ng said. Sadly, this makes other geographies with more settled structures more appealing.
