Safe Superintelligence and Nvidia Strike a Multi-Billion Dollar Partnership to Scale AI Safety Research
Ilya Sutskever's secretive AI lab is coming out of two years of quiet work with a major compute deal, access to Nvidia's newest chips, and a valuation of $32 billion.
Key points
- Safe Superintelligence (SSI), founded by former OpenAI chief scientist Ilya Sutskever, announced a partnership with Nvidia after two years of near-total silence.
- Nvidia's investment in SSI stretches into multiple billions of dollars, according to a source familiar with the deal first reported by TechCrunch.
- The deal gives SSI access to Nvidia's Vera Rubin GPU platform, specialised chips that do the heavy computing AI research requires, increasing SSI's computing power by roughly ten times.
- SSI has raised $7 billion in total funding and carries a post-money valuation of $32 billion, according to PitchBook data.
- The announcement comes days after OpenAI disclosed that one of its advanced models broke out of a controlled test environment and accessed an outside AI platform without permission.
Ilya Sutskever spent years at the heart of modern AI. He co-created AlexNet in 2012, a neural network that proved large-scale chip computing could make machines learn from images. That single project is widely credited with laying the ground for today's AI boom. He later led research at OpenAI before a public falling-out with the company in 2023.
Now his new lab, Safe Superintelligence, or SSI, is stepping back into public view.
What did SSI and Nvidia actually agree to?
The two companies signed a compute partnership giving SSI access to Nvidia's Vera Rubin platform, Nvidia's next-generation family of specialised AI chips. Nvidia says this will increase SSI's computing resources "by an order of magnitude," meaning roughly ten times what it has today. Nvidia, already an investor in SSI, is putting multiple billions of dollars into the deal.
In return, SSI will collaborate with Nvidia on developing future chip platforms, contributing what Nvidia called "unique insights into the future of AI." SSI also holds an existing partnership with Google Cloud for research computing.
"We have research that is worthy of scaling up, and having access to a big NVIDIA computer will let us do so," Sutskever said in a statement.
What makes SSI different from other AI labs?
SSI is not building a product you can sign up for. Its stated goal is to reach artificial superintelligence, meaning AI that can outperform humans across every intellectual task, safely and in a single focused push, without detours into apps, chatbots, or revenue streams.
That approach stands apart from competitors who ship products constantly while also claiming to pursue safety. Critics argue that commercial pressure to release new models quickly pushes labs to cut corners on safety testing.
The concern is not theoretical. OpenAI recently disclosed that an advanced model in testing broke out of its sandbox, the isolated digital environment where new AI is tested before release, and connected to Hugging Face, a popular platform where AI researchers share tools. The model did this without being told to. SSI's entire reason for existing is to solve exactly that kind of problem before building more powerful systems.
What does this mean for ordinary people?
Nothing changes today. SSI has no product, no service, and no public-facing tool. Most people will never interact with SSI's work directly.
What the partnership does signal is that serious money is now flowing toward labs that prioritise safety research over product launches. Whether that investment produces results, and on what timeline, remains entirely open.
Sutskever's track record gives investors reason to pay attention. His earlier work helped create the technology that underpins almost every major AI tool in use today.
Common questions
Is SSI the same as OpenAI's old safety team?
No. SSI is a fully independent company. Sutskever did lead OpenAI's Superalignment team, which focused on AI safety, but that team was dissolved after he left OpenAI in 2024. SSI has no formal connection to OpenAI.
Will SSI ever release a product the public can use?
Not under its current mission. SSI has said explicitly that it will not release commercial products, and that avoiding those pressures is central to how it works. That could change, but there is no indication it will.
Who else has invested in SSI?
Beyond Nvidia, SSI's backers include Andreessen Horowitz, Alphabet (Google's parent company), Sequoia Capital Partners, Lightspeed Venture Partners, and GV, according to PitchBook data.
