The friction between Silicon Valley's AI industry and the US defense establishment surfaced publicly this week on two fronts.
According to Forbes reporter Ty Roush, Palantir CEO Alex Karp used a heated CNBC interview on Wednesday to blast the AI industry as “effing insane.” Karp singled out the way the US military relies on AI labs for its technology, calling that dependence “effing insane,” and he panned the fees that AI companies charge businesses.
In the same interview, Forbes reports, Karp touted a deal in which Palantir works with Nvidia's Nemotron models to serve US government agencies — positioning his company as an alternative path for the government's AI needs.
Separately, GovCIO Media & Research reports that the Pentagon is eyeing AI “agents” — software that can carry out multi-step tasks on its own — as a way to slash the time it takes to approve software for military use. Defense software typically faces long review and authorization cycles before it can be deployed, and the department is exploring whether automation could speed that up.
Together, the two items sketch a defense sector actively rethinking how it sources and clears AI: a high-profile vendor arguing the current model is broken, and the Pentagon experimenting with new tools to move faster.
Why it matters: how the US military buys, builds and approves artificial intelligence will shape both national security and which companies win enormous government contracts — and this week showed that debate playing out loudly in public.