Startups

Nvidia commits $5 billion to Ilya Sutskever's Safe Superintelligence

The chipmaker's equity check, paired with early access to the Vera Rubin platform, is set to expand SSI's compute by an order of magnitude.

By Giulia Donatello · Startups & funding · 2026-07-31 · 3 min read

Photo: Unsplash / Christian Wiediger — Close-up of an Nvidia GeForce RTX graphics card

Nvidia is putting $5 billion into Safe Superintelligence, Ilya Sutskever’s product-free research lab, according to people familiar with the matter cited by Bloomberg. The joint statement issued Monday doesn’t confirm the number, but it does confirm the shape of the deal: an equity investment paired with early access to Nvidia’s next-generation Vera Rubin platform, aimed at scaling SSI’s compute “by an order of magnitude.”

The strategic logic is legible in both directions. Nvidia gets an equity position in what’s currently the most ideologically distinct frontier lab, and SSI gets guaranteed silicon at a moment when compute allocation is the binding constraint on serious research. Bloomberg’s sources say the deal came together in a matter of weeks, after Nvidia obtained what the joint release calls rare access to SSI’s closely guarded research direction, which Sutskever’s team has been quietly advancing for the last two years.

SSI raised $1 billion at a $5 billion valuation when it was founded in 2024, backed by Andreessen Horowitz, DST Global, Sequoia Capital, and Greenoaks. By February 2025 that mark had run to $32 billion on a $2 billion round. Nvidia was already on the cap table. The company also has an existing compute partnership with Google Cloud, so the Vera Rubin commitment stacks rather than replaces.

The Sutskever quotes in the joint release are worth reading closely. He calls it a “big bet on the Vera Rubin platform” and describes SSI as having produced “research that is worthy of scaling up.” That’s the language of a founder who has spent two years insisting on a “straight shot” approach, no products, no revenue cycles, and now needs to justify accepting an order-of-magnitude compute upgrade without appearing to compromise the thesis. Jensen Huang, for his part, framed Sutskever as having pioneered fundamental breakthroughs at the foundation of modern AI, beginning with AlexNet, situating the deal inside Nvidia’s own origin lore.

The subtext is elite psychology. Sutskever left OpenAI months after the failed attempt to oust Sam Altman, taking the Superalignment agenda with him into a lab that refused to ship. The industry’s largest hardware vendor has now written him the biggest check of his career to keep not shipping.

Sources