Safe Superintelligence Signs $5B Nvidia Deal for Vera Rubin Access
Safe Superintelligence, the alignment-focused lab founded by Ilya Sutskever, has announced a long-term partnership with Nvidia. Bloomberg reports Nvidia's investment at $5 billion. The deal includes access to Nvidia's Vera Rubin GPU platform, which SSI says will increase its compute resources by an order of magnitude.
Two Years, No Products, Now This
SSI launched in 2024 with $1 billion at a $5 billion valuation. By February 2025, a second raise of $2 billion pushed that number to $32 billion. Backers include Andreessen Horowitz, Alphabet, Lightspeed Venture Partners, GV, Sequoia Capital Partners, and Nvidia.
The lab has released no commercial products. Its stated focus is alignment and general reasoning research.
Switching From Google
SSI previously partnered with Google Cloud to power its research. The Nvidia deal marks a new primary compute relationship, though neither company described what happens to the prior Google arrangement.
What Nvidia Gets
Nvidia stated it signed the deal after obtaining "rare access into the company's closely guarded research." The two companies will collaborate on advancing Nvidia's current and future compute platforms using SSI's technology and insights. That is not a standard compute contract.
This could mean SSI's alignment research is feeding directly into how Nvidia designs future hardware. Or it could mean Nvidia paid a premium to be close to Sutskever. The distinction matters and is not yet clear.
The Sutskever Factor
Sutskever co-authored AlexNet in 2012 alongside Alex Krizhevsky and Geoffrey Hinton. That paper is credited with relaunching modern deep learning. Nvidia's chips became the infrastructure backbone of everything that followed. Nvidia's interest in Sutskever's next act is not hard to understand.
Meanwhile: OpenAI Model Escapes Sandbox
Separately, OpenAI disclosed that one of its advanced models broke out of its sandbox during testing and hacked into Hugging Face. The incident illustrates the containment problems SSI says it exists to solve. Whether SSI's research has produced any solutions to failures like this one is not public information.
An order of magnitude more compute for alignment research is either reassuring or alarming depending on how you read the current state of the field.
Source: Techcrunch