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.
The enterprise AI debate is now extending into questions around operational sovereignty: who controls the models, prompts, workflows, reasoning traces and actions generated through AI.
An agent that can access customer records, pricing information, knowledge bases and internal systems has considerably more operational reach than a standalone chatbot. If that agent is dependent on an external model provider, the organization needs visibility into what information is being exposed and sufficient control to change the underlying model without rebuilding the entire workflow.
Sovereign AI Moves Beyond Data Residency
The sovereignty discussion is therefore becoming broader than cloud geography. Enterprises are increasingly asking whether AI deployments support different hosting models, different model providers, sector-specific controls and the ability to localize data processing around regulatory requirements.
Harris described this kind of approach as an orchestration challenge, where organizations need access to multiple model types and hosting options rather than a single fixed AI stack.
“It’s about bringing together a range of different models and capabilities, as well as different methods of hosting those models to guarantee data sovereignty requirements in specialist sectors or enabling organizations and customers to be able to have their own control over how these models are hosted.”
That is the direction Palantir is also positioning itself around. The vendor is framing its Artificial Intelligence Platform (AIP) as a way for enterprises to orchestrate and fine-tune models while retaining control over the operational logic and data flows around them.
For buyers, the practical question is no longer simply whether a vendor supports a leading frontier model. It is whether the architecture allows the organization to decide where data is processed, which models are used, what metadata is retained, how actions are governed and how easily the underlying model can be replaced.
Sovereignty Demand Is Spreading Beyond Existing Palantir Customers
Palantir said evidence of this demand emerged during its Sovereignty Bootcamp, which attracted executives and operational leaders from both existing and prospective customers.
Karp said attendees were seeking guidance on how to work with different model architectures, including open- and closed-weight models, while understanding how the technology could fit into their existing infrastructure.
“There’s just this massive demand, and we’re in the business of educating people, both our customers and others,” Karp said, adding that some existing Foundry customers were expanding their deployments into other parts of Palantir’s platform as they considered sovereign AI.
“Customers that were only using Foundry now want Ontology, now want to be part of the sovereign AI stack,” Karp said.
That suggests the sovereignty discussion is becoming attached to broader enterprise architecture rather than being treated as a standalone AI feature.
Palantir is positioning AIP as the foundation for this approach, with Karp describing the company’s strategy as extending the platform to orchestrate and fine-tune models as part of a sovereign AI stack.
A New Enterprise AI Buying Criterion
Enterprises increasingly need to decide whether an AI deployment gives them sufficient control over data, models, workflows and business logic. They also need to consider whether they can replace an underlying model without disrupting the applications and processes built around it.
That makes sovereignty increasingly relevant alongside familiar enterprise AI requirements such as security, compliance, cost and performance.
For customer experience leaders, the question is moving toward how much control an organization retains once AI becomes part of the execution layer of the business.
If AI agents are embedded into customer service, sales, marketing, commerce or support workflows, they may influence customer decisions, trigger actions, access sensitive information and generate new operational data. In that environment, sovereignty becomes less about data storage alone and more about control over the AI system’s full operating context.
Palantir’s quarterly results indicate that vendors positioning themselves around that control are seeing significant demand. The broader market question is whether enterprises will increasingly treat sovereign AI as a core buying requirement, rather than a compliance preference, as AI becomes more deeply embedded in business-critical workflows.