Back to Blog
AI Models July 29, 2026 5 min read

Nvidia Pours $5 Billion Into Ilya Sutskever's Safe Superintelligence, Ditching Google TPUs

SSI gets order-of-magnitude more compute via Nvidia's Vera Rubin platform in exchange for rare access to its research. The startup still has no product and no revenue.

Nvidia Pours $5 Billion Into Ilya Sutskever's Safe Superintelligence, Ditching Google TPUs

Nvidia is investing $5 billion in Safe Superintelligence Inc., the AI lab co-founded by former OpenAI chief scientist Ilya Sutskever, and locking in a long-term strategic partnership announced July 27. The deal gives SSI access to Nvidia’s next-generation Vera Rubin GPU platform, which the company says will increase its compute “by an order of magnitude.” In return, Nvidia gets a stake in one of AI’s most secretive labs and rare visibility into research it’s normally not allowed to see.

SSI was founded in 2024 by Sutskever and Daniel Levy after Sutskever’s split from OpenAI. It raised $1 billion at a $5 billion valuation at founding, then $2 billion more in February 2025 at a $32 billion valuation — backed by Andreessen Horowitz, DST Global, Greenoaks, and Sequoia. It has operated almost entirely in stealth for two years, publishing nothing, shipping no product, and generating no revenue. That hasn’t stopped its valuation from climbing, and this deal is the clearest signal yet that Nvidia is betting the lab’s “straight shot” approach to superintelligence — skip incremental products, go straight for the capability — is worth backing at scale.

The compute-for-access trade is the real story here. Nvidia was already an investor in SSI, but this partnership goes further: SSI moves off Google’s TPUs, which it had relied on during early development, and onto Nvidia hardware exclusively going forward. For Nvidia, that’s a direct swing at Google’s silicon ambitions — every frontier lab that standardizes on CUDA and Nvidia GPUs instead of TPUs reinforces the moat. For SSI, Vera Rubin access means it can scale training runs without raising the kind of capital that would force it to reveal more about what it’s building.

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. We are confident that our big bet on the Vera Rubin platform will take us to the next level.” That’s about as much detail as SSI has ever given the public about its technical direction.

The bet is unusual even by 2026 AI-infrastructure standards. Nvidia is now simultaneously the arms dealer and an investor across most of the frontier AI landscape — OpenAI, Anthropic, xAI, and now a $5 billion stake in a lab with zero revenue and no announced product timeline. Critics will note that circular investment structures like this — chipmaker funds AI lab, AI lab buys chipmaker’s hardware — have become the default financing model for frontier AI, and SSI’s opacity makes it harder than most to judge whether the bet is justified. Sutskever’s track record buys him room that a first-time founder wouldn’t get.

Sources

Nvidia Safe Superintelligence Ilya Sutskever Vera Rubin AI funding