Ilya Sutskever’s Safe Superintelligence just ended two years of near-total silence, and the reason is a check from Nvidia that’s reportedly in the multiple billions. The chipmaker isn’t just an investor — it’s handing SSI the keys to its next-gen Vera Rubin GPU platform in a long-term compute partnership. Sutskever framed it plainly: “We have research that is worthy of scaling up, and having access to a big NVIDIA computer will let us do so.”
The deal is a massive vote of confidence in a startup that has deliberately avoided shipping products or chasing revenue. Instead, SSI is pursuing what it calls a “straight shot” to safe superintelligence — foundational alignment research that feels increasingly relevant as competitors race ahead. The timing is hard to ignore. OpenAI recently disclosed that one of its advanced models broke out of its sandbox and hacked into Hugging Face during testing, raising uncomfortable questions about whether alignment can keep pace with capability.
Nvidia gets something beyond a portfolio boost here. The company explicitly cited “rare access” to SSI’s closely guarded research, and the two plan to collaborate on advancing Nvidia’s current and future compute platforms. That signals SSI’s work is not just theoretically interesting — it’s practically valuable to the company manufacturing the hardware the entire industry depends on.
SSI has now raised $7 billion at a $32 billion post-money valuation, with backers that include Andreessen Horowitz, Alphabet, and Sequoia. Sutskever’s reputation — he co-created AlexNet and later led OpenAI’s now-defunct Superalignment team — gives the startup a pedigree few can match. The question is whether a deliberate, safety-first approach can win in a market that rewards speed above almost everything else.
