Nvidia invests $5B in Safe Superintelligence, grants Vera Rubin compute access

Nvidia has committed $5 billion in equity to Safe Superintelligence Inc. (SSI), Ilya Sutskever's secretive AI safety lab, in a long-term strategic partnership announced July 27. The deal grants SSI priority access to Nvidia's next-generation Vera Rubin GPU platform, increasing the startup's available compute capacity by an order of magnitude over the next 12 months. In return, Nvidia secures rare access to SSI's proprietary research and collaboration rights on future compute platforms.

SSI was previously valued at $32 billion after raising $2 billion in February 2025 from backers including Andreessen Horowitz, Alphabet, Lightspeed Venture Partners, and Sequoia Capital. The startup operates under a "straight-shot" mandate: build safe superintelligence as the sole goal without commercial product distractions, much like Sutskever's stated frustration with OpenAI's pivot to commercialization. The company has released no products, models, or research papers in its two years of operation despite employing dozens of AI researchers.

For practitioners scaling AI infrastructure, the deal signals three things: frontier AI labs are still raising at mega-scale despite unproven results; Nvidia is hedging against custom silicon and alternative compute platforms by embedding deep into research partnerships; and compute capacity for safety-focused research is now moving from cloud TPU infrastructure (Google) to dedicated Vera Rubin systems. This positions SSI to scale research without the product-release pressure that defines OpenAI and Anthropic.