Answer: Safe Superintelligence and Nvidia announced a long-term partnership that gives SSI Vera Rubin access and is expected to increase its compute by an order of magnitude. Nvidia called its investment substantial but did not disclose terms; Reuters, citing a person briefed on the deal, reported a $5 billion equity investment. The companies will also collaborate on future compute platforms.
What did Safe Superintelligence and Nvidia announce?
Safe Superintelligence, founded in 2024 and led by OpenAI co-founder Ilya Sutskever and Daniel Levy, entered a long-term strategic partnership with Nvidia. The official announcement confirms a substantial Nvidia investment, Vera Rubin access, an expected order-of-magnitude increase in SSI’s compute, and joint work on Nvidia’s current and future compute platforms.
Sutskever said SSI has “research that is worthy of scaling up” and that access to a large Nvidia computer would let the lab do so. Nvidia said it entered the partnership after receiving rare access to SSI’s closely guarded research, but the companies did not publish the research milestones Nvidia reviewed.
How much is Nvidia investing in SSI?
The companies officially described Nvidia’s investment as substantial without publishing financial terms. Reuters reported a $5 billion equity investment, citing a person briefed on the deal. TechCrunch separately reported that the investment extended into multiple billions and cited Bloomberg’s $5 billion figure.
| Deal element | Official disclosure | Attributed reporting or open question |
|---|---|---|
| Investment | Nvidia made a substantial investment in SSI | Reuters reported $5 billion, citing a person briefed on the deal |
| Compute scale | Expected to increase SSI’s compute by an order of magnitude | System count and baseline capacity were not disclosed |
| Hardware | Access to Nvidia’s Vera Rubin platform | Deployment timing and configuration were not disclosed |
| Engineering | Collaboration on current and future Nvidia compute platforms | Specific joint-development deliverables were not disclosed |
| Research review | Nvidia said it obtained rare access to SSI’s research | Reviewed milestones and measurable alignment results were not disclosed |
What does an order-of-magnitude compute increase mean for SSI?
Nvidia says the partnership will increase SSI’s compute by an order of magnitude, meaning roughly tenfold relative to its previous level. Neither company disclosed the baseline, system count, delivery schedule, workload mix, training budget, or experiment throughput, so the announcement does not support a more precise capacity estimate.
More compute can allow more or larger experiments, but compute growth alone does not demonstrate an alignment breakthrough. The evidence available publicly establishes an infrastructure expansion and the companies’ intent to scale SSI’s research; it does not establish a safety result.
Why does Vera Rubin matter to SSI’s research plan?
Vera Rubin is the Nvidia platform named in both companies’ announcement as the basis for SSI’s next compute expansion. Nvidia CEO Jensen Huang said he expects SSI to pursue new breakthroughs on the platform, while Sutskever described access to a large Nvidia computer as the resource needed to scale the lab’s current research direction.
The partnership also creates a two-way engineering relationship: SSI receives infrastructure and Nvidia says it will use SSI’s insights while advancing current and future compute platforms. The announcement does not specify product deliverables, benchmark targets, or whether SSI must use Nvidia hardware exclusively.
How does Google Cloud fit with the Nvidia partnership?
SSI’s Nvidia partnership follows a 2025 Google Cloud arrangement under which SSI used Google’s TPU chips for research; TechCrunch reported that Google Cloud was SSI’s primary compute provider at that time. The Nvidia partnership report confirms that earlier Google relationship but does not disclose whether either arrangement is exclusive.
The two relationships show that SSI has publicly identified access to both Google’s TPU infrastructure and Nvidia’s Vera Rubin platform at different stages. They do not, by themselves, establish how workloads are divided, which provider carries more capacity now, or how portable SSI’s research stack is between the platforms.
What is Safe Superintelligence?
SSI was founded in 2024. TechCrunch reported that it raised $1 billion at a $5 billion valuation in 2024 and $2 billion at a $32 billion valuation in February 2025.
The research company is led by Ilya Sutskever and Daniel Levy. Its official description states one goal and one product: a safe superintelligence. Nvidia’s announcement names Andreessen Horowitz, DST Global, Greenoaks, and Sequoia Capital among SSI’s investors. The two reported rounds precede the Nvidia transaction and correct the original article’s unsupported statement that SSI emerged from stealth with $7 billion in funding.
Sutskever previously co-founded OpenAI and contributed to AlexNet, AlphaGo, sequence-to-sequence learning, and GPT research. SSI says it spent two years quietly developing a new research direction aimed at powerful, robustly aligned AI before announcing the Nvidia partnership.
Which SSI-Nvidia deal terms remain undisclosed?
The public announcement does not disclose the investment terms, an exclusivity clause, the quantity or deployment schedule of Vera Rubin systems, the research milestones Nvidia reviewed, or measurable alignment results. Those gaps limit any claim about valuation, hardware lock-in, research progress, or safety outcomes.
Operators and investors should distinguish three evidence levels: official partnership terms, financial figures attributed to unnamed deal sources, and questions the companies have not answered. The original partnership coverage remains useful for funding history and reporting context, while Nvidia’s announcement is the primary source for the disclosed technical relationship.
About the Author
