Enterprise demand for sovereign AI is accelerating, according to data analytics software vendor Palantir, as businesses become increasingly concerned about where proprietary data goes, how AI models learn from it, and what happens when critical applications become dependent on a single model provider.
Sovereign AI was a key theme of the vendor’s second-quarter earnings call, with Palantir linking the growing demand to a broader shift in how enterprises are approaching AI deployment.
Palantir reported revenue of $1.9BN for the second quarter of 2026, up 93 percent year over year, and raised its full-year revenue guidance to between $8.15BN and $8.16BN.
During the company's quarterly earnings call, Ryan Taylor, Palantir’s Chief Revenue and Legal Officer, attributed the commercial momentum in part to a change in what enterprises expect from AI.
“What enterprises demand is AI sovereignty, owning the operational definition of the data, logic, actions, and security of their enterprise.”
That demand extends beyond simply keeping sensitive data within a particular cloud or geographic region. For Palantir, sovereign AI means giving enterprises greater ownership and control across the AI stack, including the models they use, the data and metadata generated through AI interactions, the logic governing applications, and the actions those systems can take.
Enterprises Are Questioning the AI Bargain
There is a growing tension between the convenience of accessing powerful foundation models through external providers and the strategic value of the information enterprises feed into those systems.
Taylor argued that organizations risk giving away proprietary information while paying for AI services that are difficult to connect directly to measurable business outcomes. “Worse, companies are paying to give away their most important secrets, the very basis for their competitive advantage."
The more deeply AI becomes embedded in customer service, supply chains, financial operations, manufacturing or other business-critical workflows, the more important questions around control and dependency become.
An enterprise may want to use the best available model for a particular task while retaining the ability to change providers as models, pricing, policies or availability change.
That issue came up directly during Palantir’s earnings call. Palantir CEO Alex Karp said the company already provides customers with the ability to switch models, arguing that enterprises should avoid becoming dependent on a single provider.
This introduces another dimension to the sovereignty debate for technology buyers, as model portability becomes part of operational resilience.
If a model is withdrawn, restricted, materially changed or becomes commercially unattractive, organizations need a way to maintain the applications built around it.
Rhys Harris, Product Director of AI at Content Guru, told CX Today in a recent interview that organizations need to consider model flexibility, regional requirements and governance before committing to an AI architecture.
“Ultimately, you need to be conscious of working with an organization that can provide different AI models or different products with AI solutions underneath that can operate across multiple regions for the sake of data governance.”
That view aligns with Palantir’s argument that sovereign AI is not only about where data sits, but also about whether enterprises can control the operating environment around AI.
The Value of the AI Stack Is Expanding
Palantir CTO Shyam Sankar argued that enterprises are also beginning to recognize that the valuable information generated through AI extends beyond the underlying corporate data.
As organizations use AI systems, they generate metadata, reasoning traces, usage information and other operational signals. Controlling those assets could become increasingly important as AI becomes embedded into business processes.
“That’s actually probably more valuable than just the data in my enterprise,” Sankar said. “That’s been a clarion call for the market over the last quarter, two quarters.”
The first phase of the discussion around data sovereignty focused heavily on data residency and regulatory requirements, including where data is stored, which jurisdiction governs it and whether sensitive information can cross national borders.