"U.S. Must Block Advanced Semiconductor Exports to China for as Long as Possible," House Intelligence Committee's Top Democrat Warns

Speaking at a Brookings Institution event

During data center construction in the Middle East,

Guard against technology leaks involving Nvidia chips and other semiconductors.

Rep. Jim Himes, the top Democrat on the U.S. House Intelligence Committee, said semiconductor export controls should remain in place for as long as possible to slow China's pursuit of advanced artificial intelligence (AI) technology. He particularly stressed the need to guard against the risk of technology leaking to China as advanced U.S.-made semiconductors, including those from Nvidia, are supplied for large data centers being built in the Middle East.

"U.S. Must Exercise Caution as AI Data Centers Are Built in the Middle East"

Logo of Chinese memory semiconductor company ChangXin Memory Technologies (CXMT). Reuters-Yonhap

At an event on October 21 (local time) hosted by the Brookings Institution, a Washington, D.C.-based think tank, Rep. Jim Himes said, "I don't think the United States can maintain its advantage over China in the most advanced AI models forever," adding, "But we should preserve that advantage for as long as we can." He continued, "The only real tool available to us to do that is to slow the flow of advanced semiconductors into China."

Rep. Himes is one of the "Gang of Eight," a group of eight congressional leaders briefed on the U.S.'s most sensitive national security secrets. The U.S. administration may provide limited briefings on sensitive intelligence activities to these eight members rather than to the full membership of the congressional intelligence committees.

He said U.S. companies must manage the risk of advanced semiconductors being diverted as they build large AI data centers in the Middle East. He stressed that it is necessary to track where the chips are being used, particularly when Nvidia H200 chips or more powerful semiconductors are deployed.

Rep. Himes said managing the movement of advanced semiconductors is more complicated than tracking semiconductor manufacturing equipment, but that it can be done. "We can do this. But we have to be careful," he said.

"U.S. Faces Power Shortages, While China Lacks Advanced Semiconductors"

The event also addressed the constraints facing the United States and China in the AI race. Kyle Chan, a researcher at the Brookings Institution, said limited power supply in the United States and a shortage of advanced semiconductors in China are obstacles to the development of their respective AI industries.

"In the United States, the challenge is securing enough power to run data centers and semiconductor facilities and connecting them to the grid. In China, there are constant complaints about shortages of computing capacity and semiconductors," Chan said.

He explained that China is trying to make up for its disadvantage in semiconductor technology by leveraging its vast power generation capacity. This involves connecting large numbers of semiconductors with relatively lower performance and power efficiency to boost AI computing capacity.

"China is trying to compensate for its semiconductor shortage with abundant electricity," Chan said. He noted that China's annual electricity generation is more than twice that of the United States, while its power generation capacity is more than three times as large. He also said China adds more than 500 gigawatts (GW) of power generation capacity each year and plans to invest more than $700 billion in its power grid by 2030.

"China's Lead in Power Generation Does Not Make It Easy to Overtake the U.S. in the Short Term"

A 50-megawatt (MW) molten-salt solar thermal power plant in China's northwestern Xinjiang Uygur Autonomous Region in March 2025. Xinhua-Yonhap

However, he said China would not easily catch up with the United States in AI competitiveness in the short term simply by leveraging its power generation capacity. The United States has access to advanced semiconductors with high performance and power efficiency, while China's capacity to produce its own semiconductors and its supply of chips needed to train large-scale AI models are limited.

Chan said the key question is whether China can produce enough semiconductors to meet domestic AI demand while also competing in overseas markets. "In the short term, the United States and China are not competing to build AI data centers overseas," he said. "China faces a shortage of semiconductor supply and limits on its domestic production capacity."

However, he predicted that the competitive landscape could change over the medium to long term if China's semiconductor production capacity expands. Chinese companies could offer services overseas that combine AI models, hardware, and power supply. Chan likened this to a repeat of China's overseas infrastructure expansion strategy, the Belt and Road Initiative, in the AI sector.

Rep. Himes also raised allegations that China and others are fueling opposition in the United States to data center construction. "There are many claims that China and some other countries are spreading large amounts of inflammatory content about data centers," he said, adding that he did not know whether the claims were true.

This content was produced with the assistance of AI translation services.