What is Safe Superintelligence?
Ilya Sutskever's "straight-shot" AI lab building one product — a safe superintelligence — valued at $32B with no product and no revenue.
- Category
- AI Research Lab
- Headquarters
- Palo Alto, CA (+ Tel Aviv)
- Founded
- June 2024
- Employees
- ~50
- Total funding
- ~$3B disclosed (to $6B cited)
- Valuation
- $32 billion (Apr 2025)
What is Safe Superintelligence?
Safe Superintelligence Inc. (SSI) is an American AI research lab founded in June 2024 by former OpenAI chief scientist Ilya Sutskever, alongside Daniel Gross and Daniel Levy, to pursue a single goal: building a safe superintelligence. It describes itself as "the world's first straight-shot SSI lab," with "one goal and one product" — and has reached a $32 billion valuation on roughly $3 billion of disclosed equity (some trackers cite up to $6B in total commitments) without shipping any product or generating any revenue.
SSI deliberately rejects the commercial-product playbook of OpenAI and Anthropic. Its public mission statement is blunt: "Building safe superintelligence (SSI) is the most important technical problem of our time," and the company says it approaches "safety and capabilities in tandem, as technical problems to be solved through revolutionary engineering and scientific breakthroughs." There is no API, no chatbot, no demo, and — critically — no revenue and no ARR: the stated plan is that "the first product will be the safe superintelligence, and it will not do anything else up until then." Customer and user counts are therefore zero by design; the only "users" are SSI's own researchers.
The company is run by a small, elite team — roughly 20 researchers in early 2025 growing toward 50 by mid-2025 — with no sales, marketing, or product organization. Capital is poured almost entirely into compute and top-tier research talent, with Google Cloud TPUs as its primary training infrastructure. In a widely discussed November 2025 interview, Sutskever argued the industry's "age of scaling" (roughly 2020–2025) is over and that progress now depends on new research ideas rather than more compute — the thesis SSI is built around.
Despite being pre-product, SSI is one of the most valuable private AI companies in the world. Investors including Greenoaks, Andreessen Horowitz, Sequoia, DST Global, SV Angel, Lightspeed, Alphabet, and Nvidia have backed it on the strength of Sutskever's track record (a co-inventor of foundational deep-learning work and an OpenAI co-founder) rather than any traction metrics.
What does Safe Superintelligence offer?
SSI has no commercial product. Its entire output is internal research toward a single eventual product — a safe superintelligence — so its "offering" is best described as the research domains and capabilities it is building.
- Safe superintelligence (the only planned product)· Mission
- Frontier AI research· Research
- AI alignment & safety· Research
- Superintelligence capabilities· Research
- Large-scale model training· Research
- No public API / no chatbot· Status
- No revenue / pre-product· Status
- "Straight-shot" R&D· Approach
How does Safe Superintelligence make money?
It doesn't — yet. SSI has zero revenue, no pricing, and no commercial product. It is funded entirely by venture and strategic equity capital, and intends to release nothing until it has built a safe superintelligence.
There are no pricing tiers, no per-seat plans, no usage-based metering, and no enterprise contracts to list, because SSI has deliberately insulated itself from commercial pressure. As Sutskever and his co-founders framed it at launch, the structure lets them "scale in peace," pursuing long-horizon research "free from commercial pressures" and the quarterly product cycles that shape rivals. Unlike OpenAI's ChatGPT tiers (a free plan plus paid Plus/Pro/Team/Enterprise seats) or Anthropic's Claude plans, SSI offers nothing for sale at any price.
The model is therefore pure burn against a thesis: equity capital raised from investors is spent on massive compute (primarily Google Cloud TPUs) and a small number of elite researchers, with no offsetting income. Reporting indicates an unusual capital structure in which investors can earn a capped multiple (reported around 10–20x) of their money, after which upside flows to a nonprofit-aligned entity — aligning the cap table with the mission rather than near-term returns.
Growth, for SSI, means raising successively larger rounds at higher valuations to fund more compute and talent — not customers or ARR. Its ~$5B-to-$32B valuation jump in under a year was driven entirely by investor conviction in Sutskever and the team, not by any revenue ramp; there is no public path to monetization until a safe superintelligence exists.
Who leads Safe Superintelligence?
SSI was founded by three people. After Daniel Gross departed for Meta in mid-2025, co-founder Ilya Sutskever became CEO and co-founder Daniel Levy became president.
- Ilya SutskeverCo-Founder & CEO2024–present (CEO since July 2025)Co-founder and former chief scientist of OpenAI; a foundational deep-learning researcher who now runs SSI day-to-day after Gross's exit. In Nov 2025 he publicly argued the 'age of scaling' is over and a new research era has begun.
- Daniel LevyCo-Founder & President2024–present (president since July 2025)Former OpenAI researcher; the technical team reports up through him and Sutskever after the 2025 leadership reshuffle.
- Daniel GrossCo-Founder & former CEO (departed)2024–June 2025Ex-head of Apple AI and Y Combinator partner; was CEO until Meta poached him in mid-2025 to join Meta Superintelligence Labs.
How do you contact Safe Superintelligence's leadership?
SSI publishes a single communications inbox, comms@ssi.inc, on its official contact page; it does not publish individual executive emails. The verified company email format is [department]@ssi.inc, but no personal addresses are published — route all real outreach (including to the founders) through comms@ssi.inc.
comms@ssi.incHow much funding has Safe Superintelligence raised?
SSI has raised approximately $3 billion in disclosed equity funding across two rounds, most recently a ~$2 billion round in April 2025 at a $32 billion valuation. (Several outlets and trackers cite up to $6 billion in total commitments — including compute deals — but the publicly confirmed disclosed equity total is ~$3 billion.)
Round 1 — Seed, September 2024: SSI raised $1 billion at a roughly $5 billion valuation, just three months after launch. The round was backed by Andreessen Horowitz (a16z), Sequoia Capital, DST Global, and SV Angel, and was characterized as one of the largest-ever seed-stage rounds for an AI lab with no product. No single lead was named.
Round 2 — April 2025: SSI raised approximately $2 billion at a $32 billion valuation — a roughly sixfold valuation jump in under a year. The round was led by Greenoaks Capital, which reportedly committed $500 million, with participation from Andreessen Horowitz, Lightspeed Venture Partners, DST Global, and strategic investors Alphabet (Google) and Nvidia. Some headlines (e.g., PYMNTS, Globe & Mail) framed the cumulative figure as a "$6 billion" raise, which appears to conflate equity with the Google Cloud compute commitment; the confirmed disclosed equity remains ~$3B.
There has been no disclosed down-round; SSI's valuation has only climbed, and as of mid-2026 no funding round newer than April 2025 has been publicly confirmed — the $32 billion valuation remains the latest on record. Alongside the April 2025 round, Google Cloud became SSI's primary compute provider, selling it large allocations of Google TPUs — a deal that pairs Alphabet's equity stake with an infrastructure relationship.
How did Safe Superintelligence get here?
From a one-tweet launch in June 2024 to a $32B valuation, a CEO change, and a thesis shift in under 18 months.
- June 2024SSI foundedIlya Sutskever announces Safe Superintelligence on X with co-founders Daniel Gross (CEO) and Daniel Levy; offices in Palo Alto and Tel Aviv.
- September 2024$1B seed at ~$5B valuationBacked by a16z, Sequoia, DST Global, and SV Angel — among the largest AI seed rounds ever, with no product.
- April 2025~$2B round at $32B valuationLed by Greenoaks ($500M); Alphabet and Nvidia join as strategic backers; valuation jumps ~6x in under a year.
- April 2025Google Cloud TPU partnershipGoogle Cloud becomes SSI's primary compute provider, supplying TPUs for model training.
- June 2025Meta acquisition attempt rebuffedMeta tried to acquire SSI during its AI hiring blitz; Sutskever declined, keeping the company independent.
- July 2025Leadership reshuffleCEO Daniel Gross departs for Meta Superintelligence Labs; Sutskever becomes CEO and Daniel Levy becomes president.
- November 2025'Age of scaling is over'On the Dwarkesh Podcast, Sutskever argues pre-training has hit diminishing returns and progress now needs new research ideas — the thesis underpinning SSI's straight-shot bet.
Who are Safe Superintelligence's competitors?
SSI competes for talent, compute, and the superintelligence milestone against the other frontier AI labs — though, unlike all of them, it ships no product.
- OpenAIThe lab Sutskever co-founded and left; ships ChatGPT and a full product suite, whereas SSI ships nothing until superintelligence is built.
- AnthropicAlso safety-first and founded by ex-OpenAI leaders, but commercializes aggressively via Claude and enterprise APIs.
- Google DeepMindBacked by Alphabet (also an SSI investor); pairs frontier research with massive distribution and the TPU stack SSI rents.
- xAIMusk's lab moves fast and ships Grok across the X ecosystem — the opposite of SSI's stealth, no-product doctrine.
- Meta Superintelligence LabsMeta's superintelligence effort that poached SSI's CEO Daniel Gross; a direct talent and mission rival.
- Mistral AIEurope's open-weight, sovereignty-focused challenger; ships models commercially while SSI stays pre-product.
Safe Superintelligence — frequently asked questions
