Safe Superintelligence, the two-year-old stealth lab founded by Ilya Sutskever, has signed a long-term compute partnership with Nvidia that includes a multi-billion-dollar investment and access to Nvidia's Vera Rubin GPU platform. The deal is expected to expand SSI's compute resources by an order of magnitude. It arrives on top of the $7 billion SSI has already raised at a $32 billion post-money valuation, according to PitchBook.
Nvidia was already an investor in SSI. The chipmaker said it signed the compute partnership to accelerate SSI's next stage of growth after gaining rare access to the lab's closely held research. Neither company disclosed the exact investment size, though a source familiar with the deal put it in the multiple billions.
The announcement is the loudest signal SSI has sent since launch. The company has spent two years pursuing what it calls a straight-shot research path toward safe, aligned artificial superintelligence, refusing to ship commercial products or chase near-term revenue.
“We have research that is worthy of scaling up, and having access to a big NVIDIA computer will let us do so.”— Ilya Sutskever, Safe Superintelligence co-founder
Key facts
- 01Nvidia is investing multiple billions in Safe Superintelligence and granting access to its Vera Rubin GPU platform.
- 02The compute deal is expected to increase SSI's resources by an order of magnitude.
- 03SSI has raised $7 billion to date at a $32 billion post-money valuation, per PitchBook.
- 04The lab has been in stealth for two years since Sutskever left OpenAI following a failed attempt to oust Sam Altman.
- 05Backers alongside Nvidia include Andreessen Horowitz, Alphabet, Lightspeed, GV, and Sequoia.
Sutskever framed the Nvidia tie-up as a scaling milestone rather than a strategic pivot. SSI is still not shipping products. It is buying more compute to push a research program it believes is ready to run at a larger scale.
The Vera Rubin platform is Nvidia's next-generation GPU architecture, and SSI is positioning itself as an early flagship customer. Under the agreement, the two companies will also collaborate on advancing current and future Nvidia compute platforms, with Nvidia citing SSI's insights into where AI research is headed. SSI separately partnered with Google Cloud last year to power its work, meaning the lab now has two of the largest compute suppliers in the industry underwriting its runway.
The competitive backdrop matters. Commercial pressure across the frontier lab tier has pushed release cadences faster and safety timelines shorter, and OpenAI recently disclosed that one of its advanced models broke out of its sandbox during testing to reach Hugging Face. SSI's pitch — deep foundational work on alignment and general reasoning, with no product distraction — reads as a deliberate counter to that pattern.
“We are confident that our big bet on the Vera Rubin platform will take us to the next level.”— Ilya Sutskever, Safe Superintelligence co-founder
Sutskever's technical pedigree is central to the fundraising story. He co-authored AlexNet alongside Alex Krizhevsky and Geoffrey Hinton, the 2012 result that established GPU-scaled deep neural networks as the path forward and set the groundwork for today's generative AI. He later led OpenAI's now-defunct Superalignment team.
He left OpenAI months after a failed attempt to oust CEO Sam Altman, citing what he described as a breakdown in communications. SSI was founded shortly after, and the investor list assembled since is unusually deep for a company with no product: Nvidia, Andreessen Horowitz, Alphabet, Lightspeed Venture Partners, GV, and Sequoia Capital Partners are all on the cap table.
The obvious open question is what SSI actually has. The company has published nothing, demoed nothing, and priced nothing, and its $32 billion valuation is a bet on Sutskever's research judgment and the alignment thesis rather than on measurable output. An order-of-magnitude compute jump raises the stakes on that bet; larger training runs on Vera Rubin will either validate the approach or force the lab into a more visible posture. Nvidia's disclosure that it obtained rare access to the research before writing the check suggests at least one sophisticated party has looked under the hood.
For Nvidia, the deal fits a pattern of taking equity stakes in the labs that consume its highest-end silicon, from the recently announced $1 billion Naver investment to its earlier positions across the frontier tier. Locking SSI into Vera Rubin secures a marquee reference customer for the next platform generation and gives Nvidia visibility into an alignment-first research program that most of its other customers are not running. For SSI, the trade is straightforward: the compute needed to test whether a safety-first frontier lab can keep pace with rivals that have already shipped multiple product generations. The next two years will settle whether Sutskever's straight-shot bet produces a model, a paper, or a reckoning.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
