Palantir’s Sovereignty Play: Fighting AI Labs While Wall Street Waits for Proof

Despite 85% revenue growth and surging AI defense contracts, Palantir stock remains 38% below its high as CEO Alex Karp attacks Big AI labs over 'sovereign AI' shift.

On the surface, Palantir looks like a company caught between two conflicting stories. Its latest revenue figures are staggering — $1.632 billion in the most recent quarter, growth of 85% year-on-year, the highest since its IPO, with US commercial sales surging 133% to $595 million. The market capitalization has blown past $300 billion. But the stock closed Monday at €111.20, just a hair above Friday’s €111.02, and remains 38% below last November’s 52-week high of €179.98. Since the start of the year, the shares have shed 22.29%. Something is not adding up, and the company’s CEO, Alex Karp, is making sure everyone knows where he thinks the problem lies.

Last week, Palantir released a 15-point manifesto titled Institutional Sovereignty in the Age of AI, a document that formalizes Karp’s long-running critique of the big AI laboratories. The target is clear: OpenAI and Anthropic. Karp argues that these labs build foundation models, then use that dominance to move up the stack and capture the most lucrative business — the very layer where Palantir operates. David Sacks, a former White House AI adviser, backs Karp’s argument, warning that Anthropic has already taken roles that once belonged to companies building on top of its models. Neither OpenAI nor Anthropic has responded publicly; both maintain that customer data is not used to train their models. An insider from one of the labs dismissed the whole thing as “theatrics,” suggesting Karp is primarily talking his own book.

But the manifesto is not just about picking a fight. It taps into a real shift in enterprise and government thinking. What Karp dismisses as “commodity cognition” — generic AI output from token-based models — is giving way to what the industry now calls sovereign AI: systems embedded directly into proprietary data and workflows, often running in air-gapped environments. That is precisely where Palantir’s Artificial Intelligence Platform (AIP) has moved from pilot projects to full-scale deployment. A recent partnership with Mexico’s largest insurer, GNP Seguros, illustrates the commercial expansion underway in Latin America, and analysts expect the commercial division to become Palantir’s largest revenue segment in 2026, accounting for more than half of total sales.

The defense side, meanwhile, is providing the bedrock. What began as a pilot project called Maven has now been formally integrated into the US military budget as a multi-year, multi-billion-dollar programme. That kind of planning visibility is rare for a software firm. Paired with an agreement to run Palantir’s systems on Nvidia’s Blackwell Ultra hardware in fully disconnected environments, the company is positioning itself as the operating system for institutional intelligence — whether on the battlefield or in the boardroom.

Yet the market remains unconvinced. Despite a roughly 19% recovery from the June low of €93.30, the stock still trades 3.27% below its 50-day moving average of €114.50 and 16.90% below its 200-day moving average of €133.38. The annualized volatility sits at 53.07%, with the 30-day figure at 51% and the RSI at 49 — technical signals that suggest indecision rather than conviction. Analysts on Wall Street are more optimistic: the consensus target is €160.38, implying an upside of around 44.5% from current levels. Many see Palantir’s strong performance against the Rule of 40 — balancing growth and profitability — as evidence that the operational momentum will eventually overwhelm the valuation fears.

The next quarterly report, due in the coming weeks, will be an important test. Investors will watch for fresh contract wins, especially in regulated industries and government, where the sovereignty narrative could translate into real revenue. For now, Palantir remains a stock caught between an accelerating business and a market that wants to see the payoff before it pays up.