Ilya Sutskever’s Safe Superintelligence gets a multibillion-dollar compute deal with Nvidia

Safe Superintelligence Inc., the research lab founded by former OpenAI chief scientist Ilya Sutskever, has entered a long-term strategic partnership with Nvidia that gives it access to the company’s new Vera Rubin GPU platform — a deal reported by Bloomberg to be worth roughly $5 billion, though neither company has confirmed the exact figure.

SSI has stuck to an unusual strategy since its founding two years ago: no consumer products, no near-term revenue push, just a “straight shot” at building safe superintelligent AI through foundational alignment research. That approach has not stopped the company from raising substantial capital — a $1 billion round at founding, followed by a $2 billion round roughly a year later that valued the company at $32 billion — and the new compute partnership suggests investors and infrastructure partners alike are still willing to bet heavily on a lab that has explicitly deprioritised shipping products.

Sutskever framed the deal simply: the lab has research it believes is “worthy of scaling up,” and the partnership gives it the raw computational capacity to do so an order of magnitude beyond what it had access to before.