Palantir and Nebius Partner to Build Sovereign AI Infrastructure Stack

Palantir Technologies and AI cloud company Nebius have entered a strategic partnership to provide commercial customers with a complete sovereign AI infrastructure stack. Under the agreement, Nebius will become Palantir’s preferred sovereign AI infrastructure partner, with its computing and inference services eventually brought inside Palantir’s enterprise security perimeter.

The partnership is designed for organisations that want to run AI models using their own data and infrastructure while retaining control over how those models are trained, deployed and operated. The companies will also work on bringing additional AI computing capacity online, including modular data centre deployments at sites where power is already available.

Nebius Compute Moves Inside Palantir’s AI Environment

Eligible Palantir customers will be able to deploy open AI models on Nebius infrastructure and adapt those models using their own proprietary data. The arrangement is intended to allow businesses to build models tailored to specific industries or operational requirements without handing control of their data and resulting models to an external AI provider.

Nebius provides AI-focused cloud infrastructure designed around large-scale computing workloads, while Palantir brings its software layer for data integration, security and AI deployment. Bringing the two together is intended to give customers both the computing capacity needed for advanced AI and tighter control over the environment in which those systems operate.

Sovereign AI Becomes the Core Focus

The partnership reflects growing demand from businesses and governments for AI systems that can operate under defined national, regulatory or organisational controls. Sovereign AI generally refers to keeping control over critical elements such as data, computing infrastructure, models and access within a trusted environment.

Palantir's Sovereign AI Operating System uses its Artificial Intelligence Platform, Ontology, Foundry and Apollo products to control how organisations deploy AI. The company says integrating Nebius infrastructure into this environment will extend that control to the underlying compute and inference layer.

Companies Plan Faster AI Capacity Deployment

Palantir and Nebius will also work together to increase available AI computing capacity for customers. One approach involves modular data centre deployments at locations that already have access to electricity, potentially allowing new capacity to be brought online faster than building conventional facilities from scratch.

The approach fits Nebius's broader infrastructure strategy, in which the company works with infrastructure partners that finance and own data centres while Nebius supplies the AI cloud architecture, hardware specifications and software platform. That model allows the company to expand computing capacity without having to fund every facility itself.

Open Models Take Centre Stage

The partnership is also built around the use of open AI models. Customers will be able to run and continuously adapt these models using their own data, allowing them to develop systems designed for particular domains rather than relying entirely on general-purpose closed models.

For industries handling sensitive information, this can be particularly important. Financial institutions, manufacturers, healthcare organisations and other enterprises can require strict controls over where data is processed and who can access AI systems.

Palantir and Nebius Target Growing AI Compute Demand

The deal comes as demand for AI computing continues to outpace available capacity, creating pressure on cloud providers and data centre operators to bring more GPU infrastructure online. Nebius has been expanding its AI cloud business around purpose-built computing infrastructure, while Palantir is pushing its AI software deeper into commercial and government organisations.

The companies now plan to combine those strengths. If the integration proceeds as planned, Palantir customers will gain access to Nebius's computing and inference infrastructure through Palantir's controlled environment, giving enterprises another route to deploy advanced AI without giving up ownership of their data and models.