Palantir Crushes Earnings as Karp Preaches “Sovereign AI”

THE GIST

Our analysts just identified a stock with the potential to be the next Nvidia. Tell us how you invest and we’ll show you why it’s our #1 pick. Tap here.

Palantir and its CEO, Alex Karp, have built a brand as the bad boys of AI. Even with the stock cut in half since November, the Palantirians kept faith in their brash philosopher-king, his state-of-the-art ontology, and his America-first techno-populist gospel. Now, the haters are the ones scratching their heads, with Michael Burry chief among them.

WHAT HAPPENED

Palantir crushed its Q2 ’26 earnings after the bell Monday, printing a double beat on revenue and EPS with the U.S. government still cutting checks. Revenue grew 93% year over year and 19% quarter over quarter to $1.935 billion against $1.81 billion expected. Adjusted EPS landed at $0.41 versus $0.35 expected, on cash from operations of $1.216 billion and adjusted free cash flow of $1.220 billion, both running 63% margins. Palantir is hoovering up cash from sales, spending what it needs on capex, and finishing with nearly the same pile it started with.

U.S. revenue grew 115% year over year and 23% quarter over quarter to $1.573 billion, while U.S. commercial ran hotter still at 149% year over year and 28% quarter over quarter to $764 million. The company also closed 220 deals worth at least $1 million, 98 worth at least $5 million, and 73 worth at least $10 million. Markets responded by sending the stock up more than 30% intraday Tuesday.

There’s a Florida wrinkle too. Every executive and employee who followed the company to a state with no personal income tax now holds their stock a lot more efficiently, which is a nice thing to slide across the table at a recruit weighing whether to become a Palantirian.

WHY IT MATTERS

The numbers explain some of the turnaround while Karp’s newest piece of marketing explains the rest. Oh, and he’s calling it “Sovereign AI.”

“A global movement is emerging,” Karp writes in his latest shareholder letter. “The revolution for independence and AI sovereignty is now well underway.” From there he warns about handing the people who build the language models the keys to your institutions, and about letting those models loose inside your home.

The Nvidia partnership Palantir struck in late July is that sermon with a price tag attached. Nvidia’s Nemotron models ship open-weight, so customers can download them, run them on hardware they own, and never route a single token through a frontier lab. Sovereignty as a product feature?

Karp then cops to the Marxist overtones of labor seizing the means of production back from the frontier labs, then declines to name Anthropic or OpenAI while doing it. Convenient, because Palantir has no LLM of its own and leaned on Anthropic’s Claude while working with the U.S. military in Iran, per the Washington Post. Sovereign only to a point, we guess.

One stock. Nvidia-level potential. 30M+ investors trust Moby to find it first. Get the pick. Tap here.

He’s also trying to balance American exceptionalism against Chinese open-weight models like Qwen, DeepSeek, and GLM that keep landing on par with Anthropic’s Claude Fable 5 and OpenAI’s GPT-5.6. And he breezes right past Nvidia owning roughly 90% of the AI training market on the strength of one flagship architecture. For a revolution about owning the means of production, that’s a load-bearing blind spot. Jensen Huang is not ceding his company to the proletariat over the “token industrial complex.” He’ll sell you the hardware to take the means back from somebody else.

Quasi-communist corporate cosplay aside, the shift from closed frontier models to cheap open-weight ones genuinely helps Palantir’s bottom line. Cheaper models mean cheaper inference underneath historically high-margin software, and AIP, Palantir’s Artificial Intelligence Platform, will run whatever model you point it at.

Option 1: the customer brings their own open-weight model, Palantir still charges on value created, and the savings on frontier API fees drop straight to margin. Option 2: Palantir runs the cheap model itself, passes some of the savings along, and wins accounts that were priced out before.

Either way, cheap AI widens the gross margin while Palantir keeps cosplaying as the resistance against the AI aristocracy.

Also (and we’d be neglectful not to mention it) Palantir’s stock surge put it within kissing distance of Moby’s $165 price target for Q2 2027. We remain even more confident now.

WHAT’S NEXT

Wall Street piled on after the print. Deutsche Bank upgraded PLTR to Buy with a $200 target, and Citi took its target to $245 from $200.

Which leaves Michael Burry as the least happy man in the trade. Per his latest Substack post, he’s still holding half his PLTR short.

Expect a tweet.