In Hearing, Congress Says China Stealing American AI Models, But Witnesses Focus On More Obvious Problems

In Hearing, Congress Says China Stealing American AI Models, But Witnesses Focus On More Obvious Problems

Leadership of the House Select Committee on the CCP is convinced China is stealing American AI models. That’s why China has an OpenAI rival, called DeepSeek, and the Europeans have nothing.

“They rely on Western AI models to develop their own models and AI services,” said Committee Chairman John Moolenaar (R-MI-2) in a hearing on April 16 about China’s rise in artificial intelligence and all the super computing hardware that goes along with making that happen. “To try to keep up in the AI race, Chinese companies are buying what they legally can under existing export control regimes and stealing what they cannot. This pattern of legal purchases and theft is happening at every layer of the AI technology stack,” he said in his opening remarks.

The main message from Moolenaar was that the CCP seeks full AI domination by 2030. It lacks the capital and talent to win fairly, so it steals.

Congress is increasingly worried about China’s rise up the AI food chain. This is the one segment of high tech where the U.S. has no other real rival – not Europe, not Russia, not even the tech savvy Japanese and South Koreans. The only one out there is China.

China’s smuggling of restricted, advanced AI chips – likely through the world’s largest transshipment port in Singapore – is a pervasive threat. Over the last 12 months, the Trump administration disrupted multiple smuggling rings that sought to sell China banned American-made AI chips.

In March, the Department of Justice announced a $2.5-billion chip smuggling case involving a company called Supermicro, and its co-founder Wally Liaw. He and his alleged conspirator used a hair dryer to remove Supermicro serial number labels and then put those labels on fake server boxes. They created fake AI servers to fool U.S. enforcement officers into thinking the chips were in Southeast Asia when they’d already been smuggled to China.

Worth noting, Nvidia CEO Jensen Huang, makers of graphic processing units used to build large language model AI platforms like DeepSeek, said last year that there was no evidence of smuggling. Haung is a one-man lobby shop. He has been successful at getting the Trump administration to allow for Nvidia GPUs to be sold in China, despite on-again, off-again restrictions.

The China cheat code. How do they do it? The House Select Committee on the CCP heard from witnesses who think a lot of it has to do with theft. Image pulled from witness written testimony.

Nvidia and other chip makers want to sell to China. But the AI platform giants, like OpenAI, Anthropic, and Google are less inclined. They are not allowed in the China market, of course.  They recently banned together to form the Frontier Model Forum, a joint effort to stop China’s campaign to reverse engineer their AI models.

Earlier this year, Beijing-based 360 Digital Security Group developed AI tools designed to spot security flaws for hacker exploits. Their product resembles the new AI model from Anthropic, called Mythos, which does the same thing.

But while Moolenaar talked up IP theft, or how China uses American AI models to train their AI models, the three witnesses had other concerns.

Export Controls & Beyond

Export controls are working, insofar as we have heard from China Big Tech that they were struggling to get chips and were hoping Beijing’s plans to move quickly up the chip making supply chain would help. One witness described these export controls as using your fingers to plug holes in a sinking sail boat.

“China is now racing to build a resilient domestic semiconductor industry across the entire supply chain. Photoresist, etching and deposition tools, chip design software, memory, packaging, and even advanced lithography machines. Export controls are not a panacea, but they have bought us some time. The question is how we use this time effectively,” said Kyle Chan, a Thornton China Center fellow at Brookings [Testimony].

China’s investing in their native semiconductor industry in response to expert restrictions and outright bans to companies like Huawei, for example. Witnesses agreed that these were temporary and partial measures that needed to be accompanied by other trade and investment policies.

“The details of what that looks like would have to be determined by the executive branch, but I think that we should broadcast a willingness to apply tariffs,” said Yusuf Mahmood, director of the America First Policy Institute [Testimony].

On January 14, the President issued Proclamation 11002 under Section 232 imposing a 25% tariff on a narrow subset of advanced semiconductors. Taiwanese companies were granted exemptions by the Commerce Department if they invested in the U.S., so this brings the behemoth Taiwan Semiconductor’s effect rate down to the standard MFN rate.

Still, one witness said the 232s on advanced chips was not enough.

“We have to focus on foundational chips because in every single AI system, in every single device, if you look at this iPhone, there’s only three advanced chips in there. There’s 150 foundational chips that drive this phone,” said Dmitri Alperotvitch, Executive Chairman of DC-based think tank, the Silverado Policy Accelerator [Testimony]. “China is trying to dominate that industry. We need to make sure that allied countries and the United States in particular invest in fabs across the spectrum of both advanced, foundational, logic, memory,” he said, marking one of the first times in a hearing of this kind that commodity chips were offered up as an item worth saving.

Committee member Rep. Gus Bilirakis (R-FL-12) called the scale of vulnerabilities in the domestic chips supply chain “staggering.”

Witnesses recommended the Committee advise the White House not to go too easy on China in the AI race.

“We need to hold the line on export controls and close the loopholes,” said Alperotvitch. “Providing China with cutting-edge AI chips is the modern equivalent of selling rockets to the Soviets during the moon race. We need to focus on the resiliency of the semiconductor supply chain. The efforts by this administration and the CHIPS Act that this Congress has passed are really vital to this effort,” he said.

He recommended the AI Overwatch Act – a Republican-sponsored bill that would require mandatory congressional notification before any such export is approved, plus assertions that chip exports do not come at the expense of domestic demand. And the MATCH Act – a bipartisan bill that would ultimately need support from allies like the EU, Japan and South Korea to work.

With What Will We Power All Of These Data Centers?

Those two bills are effectively Cold War 2.0 framing bills to choke China off from advanced semiconductor technologies. Viewed this way, Congress sees AI as a national security imperative in which it is us versus them. One side leads. The other must follow.

This might work, if the U.S. was not dependent on China for critical minerals. Enforcing even tougher export restrictions might also work if the U.S. was capable of building out its power grid to run the computing power needed for AI.  China can build its power grid faster than we can, Chan said.

“We lead in several critical layers of AI: new models, advanced semiconductors, and compute infrastructure. But on energy, China is ahead,” he said. “In the past four years alone, China built the equivalent of the entire U.S. power grid. Energy is one of our biggest bottlenecks, and we must find ways to add capacity while keeping a lid on energy bills for our local communities. The stronger AI stack includes energy.”

Today, various analysts predict that American AI developers are headed for an energy bottleneck that will challenge the AI infrastructure boom.

Some good news here. Four days after the Committee hearing, the White House waived certain requirements in the Defense Production Act (DPA) to streamline procedures to make it less cumbersome to expand the power grid. The announcement signals industrial policy via procurement of goods like electrical core steel, plus some flexibility and speed in how government capital for the main parts in grid infrastructure is deployed.

China’s battery makers like CATL and Gotion announced plans in the first quarter to add more than 600 GWh of new production capacity for the battery energy storage systems (ESS) market. These investments underscore global demand for energy. CATL and Gotion’s investments amount to roughly 10 times the 58GWh of total capacity installed across the U.S. as of year-end 2025.

An upcoming report by CPA senior economist Mihir Torsekar notes that U.S. imports of the three categories of electrical equipment that power data centers — transformers, switchgear, and lithium-ion batteries — grew from $33.2 billion in 2020 to $77.1 billion in 2025. Unlike the computing stack, where tariffs drove a rapid supplier shift away from China, the power stack is still heavily China dependent.

China supplied 59% of U.S. imports of lithium-ion batteries in 2025 and controls nearly all of the upstream lithium ion phosphate production.

McKinsey estimates that U.S. data center electricity demand will rise from 147 terawatt-hours in 2023 to 606 terawatt-hours by 2030, consuming nearly 12% of total U.S. power demand. That surge is colliding with grid infrastructure that was never designed to accommodate it.

Data center operators are increasingly turning to on-site battery energy storage systems to bridge the gap between what the grid can deliver and what their facilities require. That buildout is accelerating with Benchmark Mineral Intelligence saying battery storage will account for 41% of total U.S. battery demand this year, up from 26% two years earlier. China runs that market.

“U.S. policymakers at the federal and local levels must develop more streamlined permitting and licensing procedures, provide more resources for approving interconnection queue requests, and address supply chain bottlenecks in key electrical components, such as transformers and high-voltage transmission equipment,” Chan recommended.

MADE IN AMERICA.

CPA is the leading national, bipartisan organization exclusively representing domestic producers and workers across many industries and sectors of the U.S. economy.

The latest CPA news and updates, delivered every Friday.

NEWSLETTER

WATCH: WORTH FIGHTING FOR

Get the latest in CPA news, industry analysis, opinion, and updates from Team CPA.