US-China AI race takes a dangerous turn: former White House official says Washington must prepare for extreme measures if China reaches AGI first

The race between the United States and China to dominate artificial intelligence could become far more consequential if either country achieves artificial general intelligence, or AGI.

A former White House national security official has argued that Washington should begin preparing for a scenario in which China makes a sudden breakthrough and moves ahead of the US in developing AGI — including examining options ranging from intelligence operations and cyber activity to, in the most extreme circumstances, military action against Chinese AI infrastructure, according to the South China Morning Post.

The remarks were made by Jacob Stokes, deputy director of the Indo-Pacific Security Program at the Centre for a New American Security (CNAS), during an online discussion. Stokes previously served on the Obama administration's national security staff.


“Trying to think through the particulars of that will be especially important, in part because it will help policymakers … start to work backwards based on the unique nature of the technology, in the same way that in a past era, policymakers would learn about nuclear weapons and … work backwards from the science to the policy implications,” he said, reported the South China Morning Post.

The South China Morning Post reported on his comments and the recommendations contained in a recent CNAS report.

Why AGI could change the US-China rivalry

AGI remains a hypothetical technology rather than an established system. Broadly, it refers to an AI capable of performing at roughly human levels across a wide range of intellectual tasks rather than being designed for a narrow set of functions.

That possibility has become a major issue in the US-China technology competition because a sufficiently advanced AGI system could have implications far beyond consumer technology.

According to the SCMP report, Stokes' recent CNAS paper, titled “Superpowers and AGI,” examines how the arrival of AGI could reshape the strategic competition between Washington and Beijing.

His argument is that US policymakers should not wait until such a system exists before considering what its geopolitical consequences might be.

Instead, government agencies should start identifying the intelligence they would need to determine how advanced China's AI capabilities are and what steps might be available if Beijing appeared close to achieving a major breakthrough.

Military strikes included among extreme scenarios

One of the most striking elements of the proposal is the suggestion that Washington should examine the possibility of targeting Chinese data centres if they became central to an emerging AGI system.

The SCMP reported that Stokes' paper calls for the US to study diplomatic, intelligence, cyber and kinetic options that could potentially prevent China from reaching AGI before the United States.

That could include attacks against facilities used to train or operate an advanced Chinese AI system, although such action would carry an enormous risk of escalation.

Stokes has argued that US agencies would first need to determine which facilities might matter, what capabilities could be used against them and what evidence would be required to justify intervention.

The question would not simply be whether a particular data centre contains advanced computing equipment. Policymakers would also have to assess whether attacking it could prevent an AI breakthrough and whether the consequences would outweigh the potential strategic benefit.

Experts warn of the danger of escalation

The proposal has drawn a sharp warning from arms analyst William Hartung of the Project on Government Oversight.

Hartung told the South China Morning Post that attacking Chinese data centres based on their possible future role in AI development could trigger a direct conflict between the United States and China, both of which possess nuclear weapons.

“Striking Chinese data centres based on how they might or might not be used in the future would be the height of recklessness,” he said.

That concern highlights the central problem with the idea.

Unlike conventional military targets, an AI facility could have both civilian and strategic uses. Establishing whether a particular site is directly connected to an AGI programme could also be extremely difficult.

A decision to strike based on uncertain intelligence could therefore create consequences far beyond the original objective.

Could the US try to steal China's AI breakthroughs?

Military action is not the only controversial option discussed in the CNAS analysis.

Another possibility raised by Stokes is obtaining China's technological secrets through espionage.

The report argues that Washington could potentially attempt to reverse-engineer or acquire Chinese AGI technology if Beijing appeared to be approaching a decisive lead. Stokes also points to China's alleged history of intellectual-property theft as part of the argument surrounding the legitimacy of such efforts.

But there is a practical problem: the US intelligence community has not been structured for decades around systematically acquiring technological secrets in the way Stokes believes might now be necessary.

That could require a significant change in how Washington approaches intelligence collection as AI becomes a strategic technology.

The US currently holds an AI lead, but China could catch up

The immediate concern is not that China has already achieved AGI.

The SCMP reported that Epoch AI currently estimates the US has a lead of several months in frontier AI models. But Stokes warned that this advantage should not be treated as permanent.

A major technological breakthrough could rapidly alter the balance.

China's progress in robotics and other forms of embodied AI is one reason researchers and policymakers are watching its technological development closely. A breakthrough in these areas, combined with advances in computing and AI models, could potentially narrow the gap much faster than expected.

That uncertainty is central to the argument for preparing contingency plans now rather than attempting to respond after a breakthrough occurs.

AI infrastructure is becoming a strategic asset

The growing importance of data centres adds another dimension to the rivalry.

Modern AI systems depend on enormous amounts of computing power, making specialised data centres and the infrastructure supporting them increasingly important to national technological capabilities.

As the SCMP noted, AI infrastructure has already begun to feature in military calculations. The possibility that computing facilities could become targets represents a major shift from viewing data centres simply as commercial infrastructure.

But treating AI infrastructure as a potential military target also creates difficult questions about civilian casualties, economic disruption and escalation between major powers.

Trump-Xi talks add another layer to the AI race

The debate comes as US President Donald Trump and Chinese President Xi Jinping are expected to hold talks in Washington later this month, with AI governance among the issues under discussion.

The meeting comes at a time when both countries are trying to establish their position on how rapidly advancing AI should be governed and how its security risks should be managed.

Stokes reportedly described a constructive conversation between the two governments as the best-case outcome, rather than expecting an immediate breakthrough on major AI governance disputes.

That cautious outlook reflects how difficult it may be for Washington and Beijing to establish common rules while simultaneously competing for technological leadership.

The bigger question: who controls the future of AI?

The debate over AGI is no longer limited to laboratories or technology companies.

If AGI eventually becomes capable of dramatically accelerating scientific research, military planning, economic activity or technological development, the country that develops it first could gain an enormous strategic advantage.

That possibility explains why some US policymakers and security researchers are beginning to consider scenarios that would have seemed extreme only a few years ago.

But the same calculations raise another danger: efforts to prevent an AI breakthrough could themselves trigger the kind of conflict policymakers are trying to avoid.

For now, AGI remains hypothetical, and there is no certainty about when — or whether — such a system will be developed.

What is becoming increasingly clear, however, is that the US-China AI race is being viewed not merely as a contest over technology, but as a potential struggle over economic power, military capabilities and global influence. The question of how far either side should go to maintain an advantage could become one of the defining geopolitical debates of the AI era.