Nvidia continues to weave its web in the AI ecosystem.
The Jensen Huang-led group announces a “substantial” investment in Safe Superintelligence (SSI), the ultra-secretive lab founded by Ilya Sutskever, former chief scientist at OpenAI.
According to sources close to the matter cited by the Financial Times, the amount would be around $5 billion as part of a “long-term strategic partnership” between the two companies.
Expanded Access to Computing Power
Concretely, this agreement will give SSI massive access to Nvidia’s high-end GPUs, notably its latest Vera Rubin platform.
The two companies say this will allow the startup to multiply its computing power tenfold over the next twelve months. A strategic shift for SSI, which until now relied mainly on Google’s TPU chips, according to the Wall Street Journal.
“We have research that deserves to be scaled, and access to a large Nvidia computer will enable us to do so,” said Ilya Sutskever. On the Nvidia side, Jensen Huang, Nvidia’s CEO, praised a researcher who has “already been at the origin of fundamental breakthroughs in the foundations of modern AI,” saying he looks forward to discovering “the new advances that SSI will achieve thanks to our Vera Rubin platform.”
The precise financial terms have not been officially confirmed, but such commitments typically come with milestones for SSI to reach.
Ilya Sutskever…the Enigmatic Figure
Born in the USSR and raised in Israel, Ilya Sutskever made his name as a PhD student in Canada under Nobel laureate Geoffrey Hinton.
Their co-authored 2012 paper on computer vision is regarded as one of the triggers for the modern AI race, showing that scaling deep neural networks, combined with vast amounts of data and compute, yields measurable gains in intelligence.
After a stint at Google, he then joined OpenAI in 2015 as a cofounder, where his work directly contributed to the birth of ChatGPT. But his relationship with CEO Sam Altman broke abruptly in late 2023 when he voted in favor of Altman’s dismissal before leaving the company a few months after Altman’s reinstatement.
He founded Safe Superintelligence in 2024, with a single stated objective: a straight-line race toward superhuman AI that is safe for humanity — what the industry calls “safe superintelligence.”
The startup quickly raised around $2 billion from heavyweight venture capital firms such as Andreessen Horowitz, Sequoia Capital, Greenoaks, and Lightspeed Venture Partners, reaching a valuation of about $30 billion last year.
The Great Ambiguity Maintained
Yet SSI remains one of the most opaque companies in the sector: no product, no scientific publication to date, and only a few dozen employees.
Ilya Sutskever, who was among the first to theorize the famous “scaling law” (more data and compute equal more capabilities), now seems more cautious about this approach. “Now that compute is important, very important, we are in a sense back to the age of research,” he said in November in a podcast.
These arrangements, where chipmakers directly fund their own customers, are proliferating in the industry. Last week, AMD announced an investment of up to $5 billion in Anthropic. A strategy that allows chipmakers to secure massive demand while ensuring loyalty of key talent in the race toward artificial general intelligence.
