When House Energy and Commerce Chairman Brett Guthrie, R-Ky., said the U.S. “is in an AI arms race with China” at a markup in January, it was routine.
Earlier that month, at a House Ways and Means subcommittee hearing, Rep. Blake D. Moore, R-Utah, aired his worry that the words were already losing their impact.
“Saying … ‘we’ve got to win the AI race with China’ is — it’s pretty common now. It’s commonplace,” he said. “And I worry to some degree that we just keep saying it and what strategies do we have?”
And at a House Foreign Affairs Committee hearing titled “Winning the AI Arms Race Against the Chinese Communist Party,” Rep. Andy Barr, R-Ky., was preparing to declare a winner.
“Because of this, we are on the precipice of losing the AI race,” Barr said, referring to China’s construction of new coal-fired power plants.
The refrain that the U.S. is in a race with China over artificial intelligence has become a favorite of both parties, but legislators use the stakes of that race to justify very different policy positions. Experts say there’s no finish line in sight, especially as AI impacts increasingly become part of the midterm campaign debate.
With the number of session days before the election dwindling, the arguments behind AI policy bills, including a need to beat China, may have more potential for impact than the bills themselves.
A survey by the RAND American Life Panel conducted in the spring of 2025 found that 37 percent of respondents rated U.S. leadership in AI development and use as “critical.”
But that desire for U.S. leadership lives alongside public concerns about AI and its impact on the workforce.
A Quinnipiac University poll conducted in March found that 80 percent of respondents are concerned about AI; 70 percent of respondents said AI is likely to lead to fewer job opportunities, up 14 percent from last year.
Are we in an AI race with China?
Senate Commerce Chairman Ted Cruz, R-Texas, framed AI as a race with China as early as the fall of 2016.
“Ceding leadership in developing artificial intelligence to China, Russia and other foreign governments will not only place the United States at a technological disadvantage, but it could also have grave implications for national security,” Cruz, then chairman of the Commerce subcommittee overseeing AI, said at the hearing’s outset.
Since then, large language models and generative AI burst onto the scene and quickly changed workplaces and even physical landscapes as companies built data centers to support their AI models’ compute power.
The Trump administration, in its AI Action Plan released last year, said that “it is imperative that the United States and its allies win this race.”
“Whoever has the largest AI ecosystem will set global AI standards and reap broad economic and military benefits,” the plan said.
Stephan Lang is a former State Department career diplomat and U.S. ambassador for international communications and information policy who focused on Asia. Now with the Center for Strategic and International Studies in Washington, he said that AI would be a “transformative technology” for the economy and national security, as well as global influence.
“I do think it’s essential to our strategic national interests that we lead in the development of that technology,” Lang said. “And so I think the stakes are very high.”
Lang argued that exporting U.S. AI models and hardware could help spread U.S. values to those countries that adopt the American tech stack over the Chinese one.
“We have in China, a [country] that sees data and uses data as a tool for monitoring its citizens. It sees networks as a tool for controlling speech and suppressing dissent,” Lang said. “And I believe that we should want to promote U.S. technology that is based on our own approach to privacy and the role of government.”
Joe Khawam, former State Department attorney and current managing director of Legal and AI Policy for the Law Reform Institute, called the framing of AI as a race “incomplete.” The institute focuses on collaborative drafting of legislation between opposing stakeholders.
“It captures the urgency around things like compute and talent and military applications and infrastructure,” Khawam said. “But it doesn’t really sort of describe the entire scope of interest, right, in AI policies.”
Khawam acknowledged that “the competition itself is real,” but noted that the narrative is “used to unlock” other policies that need to be evaluated on their own merits, including “things like preemption over state laws … faster infrastructure buildouts, federal demand or procurement creation, allied market promotions, even … export controls and inbound investor controls.”
Partisan goals
Talking points from the White House and members of Congress show the wide variety of plans an AI race can be used to justify.
For Republicans, one of the most popular is advocating for less regulation.
“To maintain global leadership in AI, America’s private sector must be unencumbered by bureaucratic red tape,” the administration’s action plan said.
In a hearing last September focused on broadband permitting, Guthrie used a similar argument about China when talking about speeding the review process in the U.S. He described the differing speeds at which the United States and China built high-speed rail lines, adding that “We need to make sure the permitting is right.”
“We don’t want to just do what China does. I’m not suggesting that. But that’s who we’re competing with,” Guthrie said.
Guthrie made his declaration of a race at a January markup of a bill that would allow the president to waive certain environmental requirements for critical mineral refineries — part of the chip supply chain that makes AI possible.
Barr advocated for more coal production to support the energy needs of data centers in order to keep pace with China.
At a House Small Business Subcommittee on Rural Development, Energy and Supply Chains hearing in January on AI data center infrastructure, Chairman Jake Ellzey, R-Texas, said investments in “advanced energy solutions” for data centers “keep us competitive with the AI race with China.”
Export policy
Democrats have also used the argument that the U.S. is racing China on AI, often to criticize the Trump administration.
House Foreign Affairs Committee ranking member Gregory W. Meeks said at a January hearing that there “seems to be an overwhelming consensus that America must lead the AI race,” and that because of that he is “alarmed” by the administration’s policy, which has included allowing for the export of certain less-advanced AI chips to China.
“This administration is ceding our advantages in the AI race and actively undermining our national security,” Meeks, D-N.Y., said. “The administration has used our export controls, not as a vital national security tool, but as a bargaining chip in tariff negotiations with China.”
Meeks advocated for his bill that would restrict the sale of the most advanced AI chips to China.
There is bipartisan support for some protections against diversion of AI chips. Last month, the House Foreign Affairs Committee forwarded a bill that would require that certain chips have security mechanisms to verify a chip’s location in case it has been diverted from its intended export location.
Committee Chairman Brian Mast, R-Fla., said the bill was about “winning the AI arms race.”
“For us to give China any capability, any inroad, any advantage in this whatsoever is nonsensical,” Mast said.
Lang said that finding a balance between supporting U.S. business interests and staying ahead of China is not easy and has led to a “tension.”
“There are different theories and approaches for how we should pursue that, based on whether or not we want to … focus on preventing China from developing technologies or or try to get it to depend … on U.S. technologies instead,” he said.
Politics
Dan Cassino, executive director of the Fairleigh Dickinson University poll and a professor of government and politics, said that so far he hasn’t seen candidates “running hard” on AI regulation and safety.
He said that AI’s impact on the midterms is more likely to be tied to affordability concerns, including the price of energy.
“Voters, rightly or wrongly, are blaming the increase in energy prices in part on … data centers used to power artificial intelligence,” Cassino said.
He added that while traditionally an issue like data centers might split voters along party lines between those wanting more economic development and those wanting to protect the environment, energy prices have brought bipartisan ire for data centers.
The narrative of a race with China could help activate Republicans and “potentially make the data centers more palatable.” However, he said a similar trade-off of higher energy prices in return for national security has been made with regards to the war in Iran, and “voters by and large aren’t buying that.”
A poll from NBC News of 1,000 registered voters conducted in late February and early March found that 33 percent of respondents said that neither party would be good at handling AI. Some 20 percent favored Republicans to react well and 19 percent favored Democrats.
Cassino said that national security and energy are traditionally Republican issues, and that the party could find success using a similar playbook as it has with cryptocurrency in appealing to young men. But he noted voters tend to blame the president for rising prices, and that downplaying costs is “not exactly a winning argument.”
