Palantir CEO Alex Karp, speaking during a G20 fireside chat with U.S. Commerce Secretary Howard Lutnick, said artificial intelligence is now tightly bound up with national sovereignty. His point was blunt: countries need their own AI application layers, dependable access to compute and power, and a plan to bring technical talent back home under what he cast as a kind of digital patriotism.
Application layers and closed training environments
Karp said the AI application layer sits at the heart of digital sovereignty. In his telling, plugging straight into frontier models often misses on precision, and results can slip over time. A purpose-built application layer, he said, gives decision-makers better visibility and a clearer way to verify what the system is doing.
And he went further. Pair that application layer with an ontology layer, he said, and countries could train models inside closed, secure environments, keeping sensitive information and country-specific advantages from drifting to foreign tech providers. In his view, that kind of setup can also be tuned to different security tiers depending on how sensitive the job is. It would let governments use weaker domestic models or open-source models for certain tasks too, trimming reliance on big foreign tech companies.
Compute and power as the base of sovereign AI infrastructure
Karp called compute the starting point of the AI stack, from data management to model control to training. Countries that build sovereign architectures, he said, have a stronger shot at keeping control over where their technology goes.
But compute is only part of it. He tied the whole thing to energy policy as well. Large-scale computation needs stable, plentiful electricity, he said, and he pointed to France’s long-running buildout of nuclear power as a strong foundation for technology growth.
Karp says expertise matters more than funding totals
Karp said many national leaders fixate on how much money they intend to pour into technology. He pushed back hard, arguing that “the amount of money is not what matters; what matters is who controls it and who it is spent on.” If the people steering that investment do not actually understand the technology, he said, bigger budgets will not fix anything.
So his view was simple: in AI development, expert knowledge matters more than capital. Decision-makers without deep technical grounding, he said, are easier targets for commercial pressure, may write rules detached from reality, and can even expose confidential information when rolling out large language models. Karp described that as an anti-sovereignty result.
Call to bring overseas talent back home
Karp said skilled technical workers can do more than judge supply chains and security limits. They can also help governments figure out the real technology stack required for deployment.
He urged national leaders to convince top engineers and researchers living overseas to come home, serve their countries, and help shape policy. Karp said these people can make huge amounts of money in the United States, so cash alone will not do the trick. Instead, he said leaders should appeal to national identity, reminding them that “the motherland is great,” and rally what he called “actual patriots” to join national teams.
His view on Europe
On Europe, Karp said the region faces an extreme disadvantage in AI. He said Europe’s experience during the GSM era convinced many people that the European approach was better, but argued that AI is a completely different kind of innovation.
According to Karp, if Europe keeps leaning on old assumptions shaped by past success, that mindset could slow both AI development and adoption.