Safe Superintelligence Inc. (SSI)

Safe Superintelligence Inc. (SSI) is a private AI research company founded on June 19, 2024 by Ilya Sutskever, Daniel Gross, and Daniel Levy. Its founding statement defines a single objective — building safe superintelligence — with no consumer products, no API, and no near-term revenue strategy. As of the most recent reporting it has released no products and published no papers, model cards, or technical reports.

Snapshot

Valuation / Funding

Overview

SSI's explicit, sole, and stated product is safe superintelligence. The founding announcement of June 19, 2024 included no consumer products, no API, no near-term revenue strategy, and no incremental shipping cadence. The company's mission framing has drawn both substantial capital, on the argument that Sutskever is among the most credible bets in frontier AI research, and skepticism, on the argument that a company cannot remain hermetic at this scale indefinitely.

Founding

SSI was founded weeks after Sutskever's May 2024 departure from OpenAI, which followed the November 2023 board-governance crisis in which Sutskever was reported to have initially supported Sam Altman's ouster before publicly regretting his role. At OpenAI, Sutskever had co-led the Superalignment team with Jan Leike; both departed in May 2024. Leike, on departing, said that "safety culture and processes have taken a backseat to shiny products," a statement widely read as context for SSI's single-mission founding posture.

The two co-founders alongside Sutskever are Daniel Gross, a former Apple AI director and Y Combinator partner, and Daniel Levy, a former OpenAI researcher.

Mission and strategy

SSI's one-page founding statement sets out three commitments:

• Single focus: "Safe superintelligence is our one goal, our one product, and our one focus." The statement describes no side products and no management distractions from near-term commercial pressure.
• Insulation from commercial pressure: the statement frames "safety and capabilities [as] twin technical problems to be solved through revolutionary engineering and scientific breakthroughs," with a structure meant to let SSI scale safely "in peace," insulated from "short-term commercial pressures."
• Straight-line path: no stated intermediate products or incremental releases before the superintelligence goal itself.

Scientific posture and publications

SSI has published essentially nothing publicly — no papers, no model cards, no technical reports. This contrasts with Anthropic's interpretability and evaluations publication record and OpenAI's system-card and Preparedness Framework cadence, and is closer to, but more restrictive than, DeepMind's pre-AlphaFold "publish sparingly" posture.

Sutskever's prior work forms the technical lineage in which SSI operates. He is a co-author of the original scaling laws paper (Kaplan et al. 2020) and of InstructGPT, the paper establishing RLHF as the canonical post-training method, and has been a long-standing public voice on scheming, alignment faking, and superalignment.

Funding and compute

SSI raised $1 billion in September 2024 from Andreessen Horowitz, Sequoia, DST Global, SV Angel, and NFDG. Multiple follow-on rounds in 2025 were reported to reach valuations of $30 billion or more, placing SSI among the most highly valued pre-revenue companies in venture capital history.

SSI has not publicly disclosed its compute posture. Partnership discussions with Google Cloud and its TPUs were reported in 2025.

On July 27, 2026 Nvidia and SSI announced a long-term strategic partnership giving SSI access to Nvidia's Vera Rubin platform and raising the lab's compute by roughly tenfold, shifting it off its prior reliance on Google TPUs. Nvidia described its investment only as "substantial" (Source: wsj.com). Reuters and Bloomberg, citing people briefed on the deal, put the equity investment at $5 billion; the same reporting recorded SSI's last valuation as $32 billion in 2025 and noted it has released no product (Source: reuters.com; siliconrepublic.com). The arrangement places SSI within the vendor-financing and strategic-stake pattern documented in Circular Financing in AI.

People

• Ilya Sutskever — co-founder and Chief Scientist; ex-OpenAI Chief Scientist; co-author of scaling laws and InstructGPT.
• Daniel Gross — co-founder and CEO; reported leadership changes through 2025. Reports in 2025 suggested internal leadership tension and an evolving role for Gross; these remain low-confidence.
• Daniel Levy — co-founder; ex-OpenAI researcher.

SSI has reportedly hired from OpenAI, Anthropic, and Google DeepMind, a pattern discussed within the broader race-dynamics discourse on researcher mobility between labs.

Policy positions

SSI has taken no visible public policy positions. It has not signed the Seoul Frontier AI Safety Commitments, has not joined the Frontier Model Forum, and does not appear in standard lists of frontier-lab signatories to voluntary commitments. It has not published a responsible-scaling policy, an evaluation posture, or a safety framework.

Relationships

• instance-of: AI Safety Cases and Frameworks (purist variant); AI Race Dynamics
• depends-on: Scaling Laws, Training Language Models to Follow Instructions with Human Feedback (InstructGPT) (intellectual lineage via Sutskever)
• related: OpenAI (origin institution), Anthropic (safety-framing counterpart), Thinking Machines Lab (post-OpenAI founder cohort)
• supports: AI Scheming, Alignment Faking (Sutskever is a named historical contributor to scheming concern)

See also

Sutskever is referenced on OpenAI, Scaling Laws for Neural Language Models, Training Language Models to Follow Instructions with Human Feedback (InstructGPT), Sam Altman, Leopold Aschenbrenner, Situational Awareness: The Decade Ahead, Jan Leike, and Statement on AI Risk (CAIS). A dedicated Ilya Sutskever entity page is maintained separately.