In Brief
Palantir CEO Alex Karp used a CNBC appearance to argue for "AI sovereignty," the idea that companies and governments should own their AI compute, data, and models instead of renting them from OpenAI and Anthropic. SmarterX founder and CEO Paul Roetzer says the underlying concern is real, but it mostly applies to CIOs and regulated industries, not teams running standard marketing and sales use cases.
What Happened
Palantir CEO Alex Karp went on CNBC's Squawk Box and unloaded on the big AI labs, arguing that companies and governments need to own their own compute, data, and AI models instead of renting them from the likes of OpenAI and Anthropic. The appearance was pegged to a new Palantir-Nvidia deal bringing sovereign AI to governments, and it turned into a fiery case for what Karp calls AI sovereignty.
His core accusation: When a business runs on someone else's model, "these people are stealing the weights and alpha of my business." Weights are the numbers inside an AI model that encode everything it has learned, and alpha is a finance term for the proprietary edge competitors cannot easily copy. Karp claims enterprise leaders are privately "livid," convinced they are paying for tokens that don't create value (tokens are the units of text AI companies bill on), that the labs are charging roughly three times what they should, and that the models have been "irresponsibly oversold." On national security, he went further: "Are we really going to outsource the battlefield of this country to the consensus view in Silicon Valley? That is f-ing insane."
Palantir is an unusual messenger. Founded in 2003 by Peter Thiel, Karp, Joe Lonsdale, Stephen Cohen, and Nathan Gettings, the company grew out of PayPal's fraud-detection work, got early backing from the CIA's venture arm, and took its name from the seeing stones in Tolkien's Lord of the Rings. Its pitch is a product called Ontology, a software layer that sits between a raw AI model and a company's data, which Karp says keeps the model from caching your data or walking off with your IP. The company followed the interview with a nine-point sovereignty manifesto, including the lines "controlling your weights is controlling your fate" and a warning against "tokenmaxxing." In his May letter to shareholders, Karp wrote, "We stand on the walls, sentinels of the inner sanctum, against the assault of AI slop."
On Episode 224 of The Artificial Intelligence Show, SmarterX founder and CEO Paul Roetzer discussed who Karp is and how seriously business leaders should take his tirade.
The Key Numbers
$309 billion - Palantir's market cap
28% - Palantir's stock decline over the past six months
3x - Karp's estimate of how much the labs overcharge enterprises
$300-500 billion - Annual revenue of the entire SaaS industry
$5 trillion - Approximate annual wages of US knowledge workers
Why the Labs Have to Go After All Human Labor
Strip away the theatrics, and the core concern is there. "It's 100% true," says Roetzer, and the math explains why. The entire SaaS industry generates somewhere between $300 billion and $500 billion a year in revenue. U.S. knowledge workers earn roughly $5 trillion a year in wages. For the valuations the labs carry, and the money they are raising ahead of IPOs, "They have to go after all human labor," Roetzer says.
Figma is the cautionary tale. David Sacks picked up Karp's argument, writing that technical customers want "control over their compute, their models, data stack, their alpha," then pointed to Figma as proof. According to reporting from The Information that Sacks cited, Anthropic blindsided the design company, then a business partner, with the launch of Claude Design. Anthropic's chief product officer had served on Figma's board until three days before the launch, and Figma's founder said Anthropic "had not been consistently honest" with the company. Figma's stock has fallen sharply this year while Anthropic's valuation has surged.
Big companies have always played this game. Roetzer points out that Apple has done the same thing to app developers, and Meta to nearly everyone in its orbit. He also recalls Sam Altman signaling this two years ago, saying in effect that if your product doesn't get better every time OpenAI's model gets smarter, OpenAI will steamroll you. The labs are not hiding the plan.
"Big companies look and see what's working, and they either try and buy up the companies that are doing it, or they just launch a competing solution."
— Paul Roetzer, founder and CEO of SmarterX, Episode 224 of The Artificial Intelligence Show
SmarterX Take
Karp has an obvious incentive here. Palantir sells the exact layer he pitched as the fix, and his own business is built on top of the frontier models he was attacking. But motive and merit can coexist. "There is an absolute argument to own more of this, to own more of the compute, the data, the application layer they talk about, so that your prompts don't go back to the labs, and they don't steal your ideas and things like that," Roetzer says.
The key is knowing whether he is talking to you. Roetzer sees this as mainly a CIO conversation, along with highly technical CEOs in regulated and sensitive industries such as government, defense, and critical infrastructure. "The people who are just trying to spin up marketing use cases, I don't know that you need to share the same level of concern and urgency when it comes to how you're using these models," he says. Standard business use cases are relatively commoditized. The stakes change when you are protecting a battlefield instead of a campaign brief.
What to Watch
Sovereign AI is turning into a product category. Karp made his case on the back of a new Palantir-Nvidia government deal. He says Palantir is completely agnostic on models and wants more frontier and open-source options, which gives every enterprise more paths to owning its AI rather than renting it.
Every Claude Design-style launch will make Karp's argument for him. The labs will keep releasing products that compete with their own customers, and each one hands ammunition to the sovereignty camp. But Roetzer expects no letup from any of the AI labs, Anthropic included.
Who Actually Owns AI Inside Your Organization?
Karp's entire pitch is about ownership, but most companies have not answered a far more basic ownership question. According to the 2026 State of AI for Business Report, 13% of professionals say no one owns AI adoption in their organization, and with another 3% unsure, roughly one in six works somewhere with no clear AI leadership at all.
Before debating whether to own your compute, data, and models, make sure someone owns the strategy. The full report, based on 2,100+ responses from professionals across roles, functions, and industries, covers where organizations stand on adoption, ownership, governance, and tooling. Read the full report →
Mike Kaput
Mike Kaput is the Chief Content Officer at SmarterX and a leading voice on the application of AI in business. He is the co-author of Marketing Artificial Intelligence and co-host of The Artificial Intelligence Show podcast.