Safe Superintelligence Inc. tracker: funding, the Nvidia deal and a lab with no product

Safe Superintelligence Inc., known as SSI, is a two-year-old lab with one stated product, a safe superintelligence, and no public product, no beta and no reported revenue. It is also one of the best-funded startups in the world. This tracker follows SSI's money, compute, people and claims, dated, with the last update on October 1, 2026. We will revise it on each new event rather than write a new article.

What SSI says it is

The company's own site states the mission in one line: "Building safe superintelligence (SSI) is the most important technical problem of our time." It adds: "We have started the world's first straight-shot SSI lab, with one goal and one product: a safe superintelligence." It says it has offices in Palo Alto and Tel Aviv. For the term itself, see what is superintelligence.

Wikipedia dates the founding to June 19, 2024, by Ilya Sutskever, the former OpenAI chief scientist, Daniel Gross, a former Apple AI lead, and Daniel Levy, a former OpenAI researcher. Sutskever's earlier work on the problem is told in our superalignment explainer.

The tracker: dated log

• June 19, 2024: SSI announced. Three founders, two offices, one goal.
• September 2024: $1 billion raised from SV Angel, DST Global, Sequoia Capital and Andreessen Horowitz, per Wikipedia.
• March and April 2025: a round at a reported $30 billion valuation, led by Greenoaks Capital per Wikipedia. TechCrunch described the company as valued at $32 billion in July 2025. The two figures come from different reports and SSI has not published either.
• April 2025: Google Cloud supplied TPUs, per Wikipedia.
• June 29, 2025: Gross left. Sutskever became chief executive and Levy president. Sutskever wrote to staff and investors: "We have the compute, we have the team, and we know what to do. Together we will keep building safe superintelligence." Meta had pursued an acquisition, and Gross joined Meta Superintelligence Labs.
• July 27, 2026: Nvidia announced a $5 billion equity investment in a strategic partnership. Bloomberg, which broke the news, reported the deal came together within a matter of weeks and gives SSI access to Nvidia's Vera Rubin hardware.

The Nvidia deal, read closely

The headline number is $5 billion, from the company that sells most of the chips frontier labs use. Reports say it expands SSI's compute by an order of magnitude and moves the lab from Google TPUs toward Nvidia GPUs. One report said the money depends on SSI reaching milestones, with the terms undisclosed. Treat that as unconfirmed: we could not find the terms in a primary document.

Chip makers investing in customers is a pattern across the race, and it ties revenue to the buyer's fundraising. The compute side of that arrangement is the physical race we cover under compute and energy.

What SSI has shown

Nothing public. There is no model, benchmark result, system card, paper or demo. The evidence of progress is the funding and the people, which is not evidence of capability. Aggregators put total funding at roughly $2 billion before the Nvidia deal, with an estimated valuation near $30 billion. These are estimates from reports, not company statements.

Compare this with the rest of the field. Labs such as OpenAI publish model cards and are tested by outside groups, as the UK AI Security Institute does. We found none of that for SSI in the sources we opened, so no outsider can check the claim of a "straight shot" to superintelligence.

Size and revenue

Wikipedia lists about 50 employees as of July 2025 and no revenue as of March 2025. We have not found a newer headcount or any revenue figure in a primary document, so both are dated and likely stale. For a company valued in the tens of billions, that is an unusually small team, which fits the lean team SSI says it is assembling. It also means each departure, like Gross's, is large in proportion.

The claim and the hype check

SSI's claim is that a small, product-free lab can reach safe superintelligence by working straight at it, without the revenue pressure of a commercial lab. The structure is a real design choice, and it is the lab's pitch. It has two weaknesses. First, with no product it has no external test, so the safety claim is unfalsifiable for now. Second, $5 billion of compute from a chip vendor creates a commercial relationship of its own.

How we will update this page

Each entry carries a date and a source type. A figure from a company statement or a filing is marked as such; a figure from press reports citing unnamed people is marked as reported. When SSI or a major outlet corrects a number, we change the entry and add a dated correction note rather than quietly rewriting it. If SSI releases a model, this page will grow a section for independent evaluations, which is the evidence that would change our verdict.

Open questions to watch

• Will SSI publish anything: a paper, a system card, a safety framework?
• Does it sign the White House commitments logged in our accord tracker?
• What are the milestones in the Nvidia agreement, and who judges them?
• Who else leaves? Gross's exit to Meta shows how contested talent is.
• Does a first model appear, and does anyone independent test it?

Our take

SSI is a big bet on a person and a thesis, not yet on a result. The money is real and dated; the capability is unseen. Until the lab shows something an outsider can test, treat "safe superintelligence" as a mission statement. We will add entries here when SSI announces funding, hires, a model or a policy position, and mark any figure that rests on a single press report.