Palantir Technologies CEO highlights AI sovereignty as key opportunity

Palantir Technologies CEO Alex Karp has a new favorite phrase, and Wall Street is listening. The concept of “AI sovereignty,” the idea that organizations should own their data, model weights, and AI infrastructure rather than renting it from Big Tech, has become the centerpiece of Palantir’s growth story.

In its Q2 2026 earnings report, Palantir posted $1.9 billion in revenue, a 93% increase year-over-year, with profits hitting $1.1 billion. The stock surged roughly 29-30% on the news. Karp called the quarter “otherworldly,” claiming he couldn’t identify comparable businesses achieving similar growth rates.

The sovereignty manifesto

Around July 1, 2026, Palantir published a nine-point manifesto on X laying out the philosophical case for AI sovereignty. The core argument: enterprises that hand their most sensitive data and AI workflows to external frontier labs are building on rented land. Karp went further during a June CNBC interview, warning that the current model risks turning companies into “vassal states” of a handful of dominant AI providers.

He even likened the dynamics of the current AI ecosystem to “Marxism.” His specific concern centers on what he’s called “tokenmaxxing,” the usage-based pricing models that frontier labs employ, which Palantir argues create misaligned incentives where providers benefit from increased compute consumption rather than efficient outcomes for clients.

The manifesto also flagged data leakage as a tangible risk. When enterprises pipe proprietary information through third-party AI systems, the question of who ultimately controls, learns from, and benefits from that data gets uncomfortably murky.

Partnerships and the infrastructure play

On September 8-9, the company announced Nebius Group as its preferred sovereign AI infrastructure partner, a designation that positions Nebius as the go-to provider for the hardware and cloud environments underpinning Palantir’s sovereign deployments.

This builds on Palantir’s existing collaboration with Nvidia, particularly in creating secure environments for running open-weight models. Open-weight models are AI systems whose parameters are publicly available, meaning organizations can inspect, modify, and deploy them without depending on the original developer.

Palantir has also been running what it calls Sovereignty Bootcamps, intensive sessions designed to onboard enterprise clients into its sovereign AI ecosystem. The company’s AIPCon 11 event in September 2026 featured participation from both Nvidia and defense contractor L3Harris, signaling that Palantir’s sovereignty pitch is gaining traction across commercial and government sectors simultaneously.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.