NVIDIA Puts $5 Billion Into Ilya Sutskever's Safe Superintelligence
NVIDIA is investing $5 billion in Safe Superintelligence and giving it access to the Vera Rubin platform, enough to raise SSI's compute by an order of magnitude. NVIDIA says it committed after being granted rare access to research SSI has kept closed for two years.
People in This Story
Ilya Sutskever
Safe Superintelligence Inc. (SSI)
Co-founder of OpenAI and SSI, and SSI's CEO since mid-2025. Key author of AlexNet, seq2seq, GPT-2/3/4. SSI announced a roughly $5B NVIDIA partnership in July 2026.
Full profile →
Jensen Huang
NVIDIA
CEO & Co-founder of NVIDIA. Built the AI hardware backbone. NVIDIA reached $3T market cap. Queen Elizabeth Prize 2025.
Full profile →Key Takeaways
- 01
NVIDIA invests $5 billion and pairs it with access to its next-generation Vera Rubin platform, which SSI says lifts its compute by an order of magnitude.
- 02
The two companies will also collaborate on NVIDIA's current and future compute platforms, using what NVIDIA calls SSI's insights into where AI is heading.
- 03
NVIDIA states it entered the partnership after obtaining rare access to SSI's closely guarded research. SSI has shipped no product and has no revenue.
- 04
SSI was founded in 2024 and is led by Ilya Sutskever and Daniel Levy. Sutskever became CEO in mid-2025 after co-founder Daniel Gross left for Meta.
Primary Documents
What Was Announced
On July 27, 2026, NVIDIA and Safe Superintelligence Inc. announced what both call a long-term strategic partnership. NVIDIA is investing $5 billion and giving SSI access to its next-generation Vera Rubin platform. SSI says the combination raises its available compute by an order of magnitude, and the two companies will work together on NVIDIA's current and future platforms.
Why NVIDIA Did It
The detail worth pausing on is in NVIDIA's own wording: it entered the partnership after obtaining rare access to research SSI has kept closed since it was founded. SSI has released no product, published little, and has no revenue. NVIDIA is underwriting a research direction it has seen and the public has not.
Sutskever's track record is the collateral. He was a co-author on AlexNet, worked on AlphaGo and sequence-to-sequence learning, contributed to the GPT series, and led the research behind OpenAI's reasoning models.
What It Signals
- Compute is the moat, and it is being allocated by conviction. A pre-product lab getting a ten-fold compute increase says frontier capacity now moves on judgement about people, not on traction.
- The chip supplier is picking winners. NVIDIA investing in a lab that buys its hardware is a closed loop worth watching, and it is not the first such deal.
- Alignment-first framing is now fundable at scale. SSI's stated purpose is a safe superintelligence, and that thesis just attracted $5 billion without a product.
This article summarizes NVIDIA's own announcement and contemporaneous reporting.