What Is Safe Superintelligence (SSI)? Sutskever's Lab in 2026

Safe Superintelligence in five facts: confirmed or reported#

Only the first $1B is a figure SSI itself states. Every later funding number and both valuations come from press reports citing unnamed sources.

What does "straight-shot" mean? SSI's mission and approach#

The mission. SSI's home page opens with "Superintelligence is within reach" and calls building safe superintelligence "the most important technical problem of our time." It describes itself as "the world's first straight-shot SSI lab, with one goal and one product: a safe superintelligence." The name, the mission and the product roadmap are the same thing.

The approach. The company says it treats "safety and capabilities in tandem, as technical problems to be solved through revolutionary engineering and scientific breakthroughs," and plans "to advance capabilities as fast as possible while making sure our safety always remains ahead." Its stated reason for having no product is structural: "no distraction by management overhead or product cycles," and a business model that insulates safety and progress "from short-term commercial pressures."

What that means in practice. A straight-shot lab skips the usual path of shipping a chatbot or API to fund the next model. Daniel Gross put it plainly to Reuters in September 2024: the plan was "to spend a couple of years doing R&D on our product before bringing it to market." The July 2026 SSI and NVIDIA release adds one line on direction: SSI "has been quietly advancing a new research direction to unlock a powerful and robustly aligned artificial intelligence." Neither SSI nor NVIDIA has said what that direction is.

Our take. "Safety always remains ahead" is a goal, not a method. SSI has published no safety framework, threshold or evaluation that would let an outsider check it, so for now the safety claim rests on the founders' reputations.

Ilya Sutskever SSI: founders and leadership changes#

Ilya Sutskever co-founded OpenAI and was its chief scientist. Reuters reported in September 2024 that he sat on the OpenAI board that voted to remove Sam Altman in November 2023, reversed his position within days, and left OpenAI in May 2024. The SSI and NVIDIA release credits him with contributions to AlexNet, AlphaGo, sequence-to-sequence learning and the GPT models, and says he "spearheaded the research that led to reasoning models such as OpenAI o1." That is the companies' description.

Daniel Gross previously led AI work at Apple and co-ran the investment partnership NFDG with Nat Friedman (Reuters, 2024-09-04). At SSI he handled computing power and fundraising and was described as chief executive.

Daniel Levy is a former OpenAI researcher who joined as principal scientist (Reuters, 2024-09-04).

The 2025 change. On 2025-07-03 Sutskever sent a note to staff and investors, posted on SSI's updates page: Gross was "officially no longer a part of SSI" as of June 29, Sutskever was "now formally CEO," Levy was President, and "the technical team continues to report to me." Gross went to Meta: CNBC reported on 2025-06-30 that Meta had hired him and Friedman for the group that became Meta Superintelligence Labs. The same CNBC report said Meta had tried to buy SSI and was rebuffed. Sutskever's note answers that: "We are flattered by their attention but are focused on seeing our work through. We have the compute, we have the team, and we know what to do."

For Sutskever's place among other researchers, see our leaders and researchers guide.

Safe Superintelligence valuation and funding, as reported#

2024. SSI's updates page says it raised $1B from NFDG, a16z, Sequoia, DST Global and SV Angel. Reuters reported the same day that SSI "declined to share its valuation but sources close to the matter said it was valued at $5 billion," and that the company then had 10 employees.

2025. Reuters reported on 2025-04-11 that Alphabet and Nvidia had invested, that SSI "was recently valued at $32 billion in a round led by Greenoaks," and that Google's cloud arm had agreed to sell SSI access to its TPU chips. Reuters said it "could not determine the exact terms." TechCrunch later put that round at $2B; we could not confirm the amount from SSI or an investor.

2026. The joint release says NVIDIA "has additionally made an investment in SSI" and that the investment, with access to the Vera Rubin platform, will let SSI "increase its compute by an order of magnitude." SSI's own updates page says "10x." The release gives no amount. Reuters, citing "a person briefed on the deal," and Bloomberg both reported a $5 billion equity investment. The release lists SSI's investors as Andreessen Horowitz, DST Global, Greenoaks and Sequoia Capital.

Our take. If the reported figures are right, SSI has raised about $8 billion. That is our sum of press numbers, not a figure SSI has confirmed, and no valuation has been reported for the NVIDIA deal. Some sites repeat $32 billion as a current valuation; it is the April 2025 figure.

SSI offices and team#

SSI calls itself "an American company with offices in Palo Alto and Tel Aviv," and says it is "assembling a lean, cracked team of the world's best engineers and researchers." The only headcount we found is Reuters' "10 employees" in September 2024. We could not confirm a current headcount.

What SSI has released as of November 2026#

Nothing an outsider can test. SSI's home page, updates page and the NVIDIA release name no model, paper, API, demo or benchmark result, and set no release date. The three updates on its site are the 2024 raise, the 2025 leadership note and the 2026 NVIDIA partnership.

What it says it has. "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 the July 2026 release. NVIDIA says it invested "after obtaining rare access into the company's closely guarded research." Both are claims about work nobody outside the two companies has seen.

Rumours. In August 2026 several outlets repeated a claim, attributed to investor Gavin Baker, that SSI would release its first model that month. SSI made no such announcement, and we could not find the original remark.

SSI vs OpenAI, Anthropic and Meta Superintelligence Labs#

Versus OpenAI and Anthropic. Both ship models every few months and fund research with product revenue. Both publish versioned safety frameworks with named risk thresholds, which our frontier companies guide compares. SSI does the opposite on each count: no revenue, no releases, no framework. Its bet is that a lab without product cycles can reach a safer result. The cost is that outsiders have nothing to evaluate.

Versus Meta Superintelligence Labs. MSL is SSI's mirror image. It sits inside a company that serves billions of users and expects $135-145 billion of capital expenditure in 2026, per its October 2026 earnings release. It says it aims to put "personal superintelligence" into Meta's products, and it has shipped Muse Spark and Muse Wonder. The two labs are linked by people as well: Meta tried to buy SSI, per CNBC, then hired its CEO.

For other startups that use the word in their mission, see superintelligence startups to watch.