Nvidia Bets ~$5 Billion on SSI, Computing Power to Scale 10x in One Year
Taylor Wilson
Nvidia is making a strategic investment of roughly $5 billion in Safe Superintelligence Inc. (SSI), granting access to its next-generation Vera Rubin compute platform and targeting a ~10× increase in SSI's compute capacity within about a year. This means → Nvidia is backing a lab that has never shipped a single product — with both cash and its most advanced hardware.
Who is SSI, and why is Nvidia investing in a company that has released nothing?
SSI was founded in June 2024 by Ilya Sutskever with a single stated goal: build safe superintelligence. The company has published no models, no papers, and no demos to date.
Sutskever is one of modern AI's key architects. His research contributions span AlexNet, AlphaGo, the GPT family, and foundational work behind OpenAI's o1 reasoning model.
Nvidia CEO Jensen Huang framed the investment as a bet on Sutskever's personal research track record. In plain terms = Nvidia is not buying a product — it is buying the proven ability of one researcher to create breakthroughs from scratch.
What does $5 billion buy?
Nvidia will give SSI access to its next-generation Vera Rubin compute platform, aiming to scale SSI's compute by roughly 10× within about a year.
Multiple media outlets report the investment at approximately $5 billion, though neither company has officially disclosed the figure.
The two companies will also collaborate on technical development across Nvidia's current and future platforms, with SSI providing research insights on the future direction of AI. This means → the deal is not a passive financial bet — it is a deep exchange of compute for research direction.
A 5× valuation jump in seven months — on what basis?
In September 2024, SSI closed a ~$1 billion first round at a ~$5 billion valuation, backed by Andreessen Horowitz and Sequoia Capital. The company had roughly 10 employees at the time.
In April 2025, SSI raised ~$2 billion more, led by Greenoaks, pushing the valuation to ~$32 billion. Nvidia and Google (Alphabet) joined as new investors alongside existing backers.
In plain terms = a company with no product, no revenue, and a skeleton crew saw its valuation leap from $5 billion to $32 billion in seven months. The market is betting Sutskever's team can produce verifiable results on the superintelligence track before anyone else.
What does a dual-compute arrangement signal?
Google Cloud had already partnered with SSI on compute. SSI is a significant customer for Google's TPUs — tensor processing units, Google's custom AI chips.
The Nvidia deal now locks in SSI's access to both of the world's most powerful AI chip ecosystems simultaneously.
This reflects a deliberate strategy: SSI is not picking sides — it is securing two parallel compute pipelines so that its research is never constrained by a single supplier.
The core question: what will prove this investment right or wrong?
SSI co-founder Sutskever announced the partnership by posting: "It's time to start Scaling." SSI said its research has reached a stage that "merits scaling up."
Nvidia said the partnership was built on being granted deep access to SSI's tightly guarded research. This means → Nvidia saw something the public has not, and made its call on that basis.
Whether SSI can use Vera Rubin's compute to push its "safe superintelligence" research to a verifiable milestone will be the ultimate test of this investment's value.
It's time to start Scaling.
Ilya Sutskever
Co-founder, SSI
(announcement on X)
Content is for reference only, not financial advice.