Palantir Taps Nebius for Sovereign AI Infrastructure Inside Its Perimeter

Palantir Technologies and Nebius Group announced a strategic partnership on September 8, 2026, under which Palantir named Nebius its preferred sovereign AI infrastructure partner and will integrate Nebius compute and inference endpoints inside the Palantir enterprise perimeter for commercial customers.

The partnership announcement, issued from Miami and Amsterdam, brings Nebius’s AI-native compute infrastructure and cloud platform to Palantir’s commercial customer base. Following an integration period, eligible Palantir customers will be able to access Nebius’s cloud and inference infrastructure from within the Palantir perimeter, giving them control over their compute, data, and models.

Compute Capacity and Deployment Mechanics

The companies said they will work together to accelerate the deployment of new compute capacity to serve Palantir customers, including through modular data-center deployments at sites where power is already available. That approach targets locations with existing power access rather than requiring new power infrastructure before capacity can come online.

Palantir said it selected Nebius because the provider was built for AI from the ground up rather than adapted from general-purpose computing, with infrastructure and software designed together for demanding AI workloads. Palantir also cited Nebius’s ability to be integrated directly into Palantir’s Sovereign AI Operating System. Palantir described Nebius as one of few providers able to make that claim.

The Sovereign AI Operating System Layer

Palantir described its Sovereign AI Operating System as built on AIP, Ontology, Foundry, and Apollo, providing the authorization and isolation layer that lets organizations train models on their own proprietary data while retaining control of their compute, models, data, and the advantage those models produce. Palantir framed the arrangement around its stated position that AI should serve an organization rather than govern it.

According to Palantir’s platform documentation, the standard Palantir architecture consists of three integrated platforms. Apollo is the continuous delivery platform that manages the underlying infrastructure hosting Foundry and AIP services, orchestrating zero-downtime upgrades across services and assets. Foundry is the foundational data operations platform, providing core capabilities for data management, logic authoring, Ontology development, analytics, and workflow development. AIP is the generative AI platform, providing secure connectivity to large language models, a development toolchain for building agents and automations, AI-enabled end-user applications, and an evaluations framework for governing AI workflows in production.

The documentation states that the integrated AIP, Foundry, and Apollo architecture is designed to function as an enterprise operating system, with a unified security architecture spanning all three platforms across infrastructure, platform, and enterprise security. At the infrastructure level, every component operates with zero trust, and at the platform level the controls include role-based, marking-based, and purpose-based access controls connected with automated lineage and auditing.

Open Models and Customer Data

Eligible Palantir customers will be able to deploy open models on Nebius infrastructure and continually adapt them using their own data for their specific domain. The companies said this gives organizations the ability to build AI that can outperform general-purpose closed models for a particular use case, while retaining control of both their data and the resulting model.

The partnership is based on a shared vision that open models and looped customer data create the smartest domain intelligence and provide better security, according to the announcement. Bringing Nebius infrastructure inside the Palantir perimeter opens that option to commercial organizations that have not previously had a sovereign option to run their own models on trusted infrastructure, the companies said.

“Nebius’ compute infrastructure powers your ability to run your own AI models under conditions you control,” said Alex Karp, co-founder and CEO of Palantir. “Our ontology and their infrastructure will undergird the sovereignty our partners are demanding.”

Arkady Volozh, founder and CEO of Nebius, said organizations need both the performance of large-scale AI infrastructure and control over their data and models. “Together with Palantir, we are bringing this to commercial clients enabling them to run their optimized open models on trusted infrastructure,” Volozh said.

Company Positions

The Sovereign AI Operating System supplies the control layer, and the partnership extends the underlying compute and inference substrate to customers that need infrastructure they can govern within their own perimeter. Palantir’s documentation states that its architecture is designed to be extended and deeply integrated with other services and applications, and that its Compute Modules framework allows developers to securely bring their own containers, including containerized large language models, into the Apollo-managed mesh.

Nebius, listed on Nasdaq and headquartered in Amsterdam, is building a full-stack AI cloud platform spanning data and model training through production deployment, serving startups and enterprises building AI products, agents, and services. The Nebius announcement confirms the preferred-partner designation and the plan to bring Nebius compute and inference endpoints inside the Palantir enterprise perimeter after the integration period, and states that the companies will work together to bring new AI capacity online faster.