Palantir has named Nebius Group its "preferred sovereign AI infrastructure partner," according to a release on Palantir's investor relations site dated September 8, 2026. The partnership aims to bring Nebius's compute infrastructure and cloud platform to Palantir's commercial customers. The announcement is about two months old.

What changed

Palantir said that, "following the integration period," it will bring Nebius compute and inference endpoints inside the Palantir enterprise perimeter. Eligible customers would then be able to use Nebius's cloud and inference infrastructure. The companies also plan to add compute capacity, "including through modular data-center deployments at sites where power is already available."

Who it is for

Palantir said the move opens the opportunity to commercial organizations "that have not previously had a sovereign option to run their own models on trusted infrastructure." The release does not define which customers count as eligible, and it names none.

How it works

Palantir describes a "Sovereign AI Operating System, built on AIP, Ontology, Foundry and Apollo." It says this system provides "the authorization and isolation layer" for training models on an organization's own data. Under the partnership, Nebius supplies the compute.

In Palantir's description, eligible customers deploy open models on Nebius infrastructure. They then "continually adapt them using their own data for their specific domain." Palantir claims this lets organizations build AI that can outperform general-purpose closed models for a particular use case.

"Our ontology and their infrastructure will undergird the sovereignty our partners are demanding," said Alex Karp, Palantir's co-founder and CEO.

Arkady Volozh, Nebius's founder and CEO, said: "Together with Palantir, we are bringing this to commercial clients enabling them to run their optimized open models on trusted infrastructure."

What is not disclosed

The release gives no commercial terms. It gives no length for the integration period, no availability stage and no go-live date. It does not say which open models will be supported.

Nebius filed the same text with the SEC as Exhibit 99.1. That exhibit lists among Nebius's risks "the ability of Nebius and Palantir to successfully market joint offerings."

Why it matters

In Foundry News's view, this extends Palantir's sovereignty pitch from the software layer down to who supplies the compute. Teams building on AIP may want to fine-tune open models on their own data. For them, the key questions remain open: who is eligible, when it ships and which models run.

The announcement makes access contingent on an integration period whose length Palantir has not disclosed. The signal to watch is Palantir documentation or release notes naming a specific Foundry, AIP or Apollo feature tied to Nebius.