Pentagon AI chief says US allies lack the resources to keep pace

The Pentagon’s most senior artificial intelligence official has said publicly what allied defence ministries have been hearing privately for the past two years: they do not have the same resources as the US, and they are unlikely to catch up anytime soon.

Cameron Stanley, the Department of War’s chief digital and artificial intelligence officer, made the comments at the Billington Cybersecurity Summit in Washington on Wednesday.

The remarks were reported by the Guardian and Inside AI News, which quoted Stanley directly.

“They don’t have the resources that we do, they don’t have the experience that we do, they don’t have the scale that we do,” Stanley said.

Stanley was referring to NATO members and the Five Eyes intelligence partners, the UK, Canada, Australia and New Zealand.

He said his office is helping these countries avoid some of the mistakes the Pentagon made while developing its own AI capabilities.

That can be seen as a straightforward offer of help. It also points to a bigger issue for America’s allies. If they rely on the US to show them how to build military AI, they are likely to end up relying on the same systems, suppliers and technical choices.

The Pentagon signed agreements with seven AI companies for classified systems earlier this year. It had already committed roughly $800mn through contracts with OpenAI, Anthropic, Google and xAI.

No European defence ministry has a budget on that scale for AI, and European countries also do not have domestic frontier AI companies comparable to the US firms they could contract with.

Stanley’s comments leave out the effort Europe is making to reduce that dependence.

European governments and defence companies have been arguing that building their own technology is important precisely because relying on American systems could become a strategic weakness.

Germany’s Helsing has raised €450mn to develop defence technology for NATO. Munich-based ARX is building land drones in the UK, while European defence startups have repeatedly called on governments to give domestic companies more support.

There is also a question about what “scale” actually means on a battlefield. Much of the US advantage in military AI comes from access to large amounts of computing power and the latest AI models.

Both depend heavily on money and procurement, rather than simply on engineering talent.

Ukraine has spent three years showing that another approach can work. Cheap autonomous systems that are improved and adapted every few weeks can have a significant impact on the battlefield.

Ukraine has a fraction of the resources available to the Five Eyes countries, but its experience has shown that large amounts of computing power are not the only way to gain an advantage.

Smaller defence budgets can also offer some advantages. A government with less money may be able to move faster, work with several suppliers and replace a system without being tied to a programme that lasts for years.

Several European defence startups are making exactly this argument, saying that software-based warfare rewards companies that can replace or improve a system quickly rather than taking a decade to develop a new one.

Whether that approach can compete with America’s advantage in computing power is still unclear. Stanley did not address that question.

The timing of his comments also matters. European governments are beginning their biggest defence spending programmes since the Cold War, and AI will be part of those decisions.

They have to choose between buying established American systems now or investing in European alternatives that may take years to reach the same level.

A senior US official saying that European allies lack the resources and scale to match America inevitably becomes part of that debate.

Even if that was not the intention, it supports the case for buying American technology rather than waiting for a European alternative.

Stanley’s comments are likely to be received differently in Brussels than in Washington. They may be accurate, but they also make a clear case for continued reliance on American AI capabilities. That is what makes them significant beyond the numbers he cited.