China Challenges U.S. AI Lead With Overlooked Advantage

China is moving aggressively to expand the data centers and power networks needed to support artificial intelligence, fueling concerns that Beijing’s ability to deliver huge infrastructure projects at speed could narrow America’s lead in the physical backbone of the AI race.

For now, the U.S. still holds a commanding numerical advantage. The Stanford Institute for Human-Centered Artificial Intelligence’s 2026 AI Index, citing Cloudscene data, identified 5,427 data centers in the United States in 2025, compared with 449 in China. Stanford noted, however, that the figures do not reflect differences in the size of facilities, their computing power or how heavily they are used.

Even so, China’s buildout is accelerating.

“We know that China is far behind, but we also know that they are moving at great speed,” China expert Gordon Chang told News Outlet.

Chang said Beijing recognizes that the U.S. advantage in AI infrastructure will be hard to overcome if American development continues at its current pace. He also warned that rising local resistance to new data center projects could hamper that growth.

President Donald Trump echoed that concern this week, cautioning that communities opposing new data center construction could be handing China an opening.

“China could not be happier with this anti Data Center movement,” Trump wrote on Truth Social. “Actually, they can’t believe it is happening!” Trump added that communities willing to welcome data centers would benefit from new jobs and increased tax revenue.

Vice President JD Vance sharpened that warning Thursday, telling Fox Business there was “a very good chance” the U.S. could lose the AI race to China if Democrats regain control of Congress.

Vance said the broader problem is America’s ability to build enough power to support data centers and other major infrastructure. “We cannot be a country that says we’re not going to build more power,” Vance said, arguing that the administration is trying to make it easier to construct power plants alongside data centers while protecting residential ratepayers from higher electricity costs.

He also pointed to the widening US-China electricity gap, saying China now generates roughly three times as much power as the United States. “We’ve got to build more power plants. We’ve got to make the regulatory state easier to work with.”

Asked about China’s rapid infrastructure expansion, the Chinese Embassy in Washington rejected the premise that AI should primarily be viewed as a geopolitical contest.

“AI should not be owned by major countries, still less dominated by contests and rivalry,” Embassy spokesperson Liu Chang told News Outlet.

Liu pointed to China’s 2025 Global AI Governance Action Plan, which calls for accelerating construction of clean power, next-generation networks, computing capacity and data centers. He said China advocates cooperation with the U.S. and wants AI development to benefit both countries and the broader international community.

Behind that diplomatic message, Beijing is undertaking a massive geographic reorganization of its computing infrastructure.

Under China’s “East Data, West Computing” initiative, Beijing is directing more computing capacity toward western regions where land is available and wind and solar resources can help reduce energy costs.

Five of China’s eight national computing hubs are located in western China, according to a June 25 report from the government-backed Digital China initiative. More than 30 data centers were under construction within a three-kilometer area in Horinger, Inner Mongolia, while intensive construction was also underway in Zhongwei, Ningxia.

That buildout caught the attention of members of the bipartisan US-China Economic and Security Review Commission during a recent fact-finding trip to China.

“I left with an intense belief that the Chinese see this as the frontier in terms of competition,” Commission Vice Chairman Mike Kuiken told News Outlet, describing AI as increasingly intertwined with robotics, biotechnology, advanced materials and manufacturing.

Kuiken said China has one significant advantage when it comes to constructing the infrastructure needed to support that technology.

“The Chinese, because they write the rules as they go, and the rule of law largely does not exist, are able to just move faster in terms of building things,” he said.

But Kuiken also identified a major Chinese weakness: access to the most advanced chips.

“The constraints that are real though, the Chinese do not have the ability to procure advanced chips,” he said, while describing Beijing’s ability to rapidly construct coal plants and other infrastructure as “quite incredible.”

The U.S. retains major advantages beyond its data-center lead. The Stanford Institute for Human-Centered Artificial Intelligence’s 2026 AI Index found that America produced 59 notable AI models in 2025 compared with 35 from China, and U.S. private AI investment reached $285.9 billion, compared with $12.4 billion in China. Stanford cautioned that private investment figures understate Beijing’s overall spending because of Chinese government-backed funds.

At the same time, Chinese models have rapidly narrowed the performance gap. Stanford’s 2026 AI Index found that as of March 2026, the top U.S. model led the top Chinese model by just 2.7%.

Chang also alleged that Beijing is seeking to exploit American opposition to data centers and claimed China-linked networks have helped support protests against projects in the US. “These protests are not entirely organic,” Chang told News Outlet.

Kuiken was cautious about portraying those local disputes as part of a broader anti-China fight, saying genuine U.S. opposition to data-center projects has also emerged over concerns including electricity costs, water use, noise and land development.

“We are building things in the U.S. faster than we have in a long time,” he said, adding that policymakers are working through infrastructure constraints. “Usually there are local issues that are driving those things.”

The emerging competition, therefore, is not simply over who develops the smartest AI model.

It is increasingly a race over who can build the electricity, data centers, chips and networks needed to run those models at enormous scale — and how quickly they can build them.