THESIGNAL

Palantir’s Blockbuster Quarter, and a Broadside at the Frontier Labs

Palantir just delivered the kind of quarter that ends arguments — and its CEO used it to start one. On Monday the company reported $1.94 billion in second-quarter revenue, up 93% year over year, with U.S. commercial revenue up a startling 149%. It raised full-year guidance above $8.15 billion. Then, in the shareholder letter, Alex Karp reached past the numbers for a philosophy fight.

“There are Marxist overtones and undertones to our business,” Karp wrote, arguing that the large-model labs “intend, knowingly or otherwise, to capture the means of production of their purported partners.” On the analyst call he was blunter still: enterprises that pipe their proprietary work into a frontier model are, he said, “paying for the right for them to migrate your IP, your know-how, your expertise to their model, so that they can build a competitive business that doesn’t require your business or people.”

The irony writes itself — AI demand drove Palantir’s record. But the argument lands because it isn’t only Karp’s. The same distrust surfaced this week from Microsoft’s Satya Nadella, and it maps onto a growing list of companies that partnered with OpenAI or Anthropic only to watch those labs launch into their category — design, healthcare ops, legal, even drug discovery. Palantir’s pitch is the mirror image: model-agnostic software that lets an organization keep control of its data and its “AI exhaust” — the prompts, orchestration, and context that quietly become someone else’s training set.

For a CTO the number that matters isn’t the 93%. It’s the strategic question underneath it: when the model is a commodity and the data is the moat, which layer of your stack are you willing to rent from a company that may compete with you next quarter? Karp’s language is theatrical; the underlying build-vs-buy calculus is not.

Markets shrugged at the rhetoric and rewarded the print: PLTR traded around $125.65, up ~2% after the release. The tell for the next four quarters won’t be Karp’s vocabulary — it’ll be whether “keep your data off the labs’ models” becomes a line item in enterprise AI RFPs, or stays a talking point.