Palantir Technologies chief executive Alex Karp warned on 3 September that falling behind in the artificial intelligence race presents a graver threat to countries than the technology's own potential harms, as governments worldwide race to secure computing capacity, electricity supply, and sovereign control over their data.

The Full AI Stack Nations Must Build

Speaking at a fireside chat alongside US Commerce Secretary Howard Lutnick in Chapel Hill, North Carolina, Karp laid out what he called the complete infrastructure stack countries need to remain competitive. 'You need access to compute, you need access to your own data,' he said. 'So the stack is compute data, ability to manage data, ability to control the models, ability to train the models, and ability to compete against certain competitors and adversaries.'

Karp noted that nations could pursue computing capacity through multiple routes, including partnerships with chipmakers such as Nvidia. He described Nvidia chief executive Jensen Huang as 'the world's best' in the field, adding that Palantir itself was exploring sovereign computing structures as an alternative pathway.

The Electricity Problem No Leader Can Ignore

Karp stressed that scaling AI infrastructure would demand substantially more electricity than most governments have planned for. 'You are gonna need a lot of electricity. How are you gonna get it?' he said, pointing to France's sustained reliance on nuclear power as a model for national energy strategy. He argued that national leaders would have to publicly account for how they intended to power data centres and related infrastructure.

This comes amid a broader global trend: technology companies and governments are already examining nuclear, renewable, and other firm-power sources to meet the surging energy appetite of AI data centres, which require continuous electricity for servers and cooling systems.

Rejecting Both Doom and Blind Optimism

Karp acknowledged that AI carries real costs and disruption, and said he did not consider the technology to be entirely beneficial. However, he argued that refusing adoption would compound the risk. 'I don't think the AI revolution is 100 per cent good,' he said. 'And with all these technologies, but you, but, but then being left behind is probably the biggest danger now.'

He also pushed back against what he characterised as exaggerated doom narratives, arguing they had frightened the public and created political obstacles to adoption. 'I don't believe in the doom message,' Karp said, adding that such warnings contained 'elements of truth' but had been amplified to the point of causing 'significant political problems.' He did not name the company he was referring to.

Who Leads in Production, Who Leads in Adoption

Karp drew a sharp distinction between building AI and persuading societies to use it. The United States, he argued, holds a commanding advantage in production. 'We have an enormous advantage of production,' he said. 'The dissemination of AI and getting people to adopt without political insanity, that's where we are.'

By contrast, he suggested that countries such as Japan, South Korea, and China may face fewer political barriers to adoption. European nations, he warned, risk serious disadvantage if regulation and public opposition slow deployment. 'The technology's not egalitarian,' Karp said. 'There's no getting around these issues, and so we have to figure out ways to bring people along and explain they'll participate.'

Founded in 2003 by Karp, Peter Thiel, and others, Palantir Technologies is headquartered in Denver and develops data integration and AI software used by government agencies, defence organisations, and commercial enterprises. As the AI sovereignty debate intensifies globally, Karp's remarks are likely to sharpen pressure on governments that have yet to define a coherent national AI infrastructure strategy.