Who Is Ilya Sutskever, the CEO of Safe Superintelligence?

Ilya Sutskever is an artificial intelligence (AI) researcher who serves as the chief executive officer (CEO) of Safe Superintelligence Inc. (SSI).

Sutskever co-founded OpenAI in 2015 as research director, later chief scientist, after co-creating AlexNet and co-developing sequence-to-sequence models.

As of October 2026, SSI lists no product, and its July 2026 NVIDIA partnership is meant to scale its compute by 10x.

What Ilya Sutskever Is Doing Now

Sutskever has been formally CEO of SSI since July 3, 2025, per an official leadership update published on the SSI updates page. The technical team at SSI reports to Sutskever.

Sutskever left OpenAI in May 2024. SSI says it handles "safety and capabilities in tandem."

• Current role: Chief Executive Officer (CEO) of Safe Superintelligence Inc. (SSI) since July 2025.
• Organization: Safe Superintelligence Inc., maintaining operational offices in Palo Alto and Tel Aviv.
• Known for: Co-creating AlexNet, co-developing sequence-to-sequence models, and co-founding OpenAI.
• Education: Honours Bachelor of Science in Mathematics (2005) and PhD in Computer Science (2013) from the University of Toronto.
• In the news for: SSI's long-term NVIDIA partnership, announced July 27, 2026, which includes an NVIDIA investment of undisclosed size.

Founding of Safe Superintelligence Inc.

Ilya Sutskever announced the founding of Safe Superintelligence Inc. on June 19, 2024, alongside co-founders Daniel Gross and Daniel Levy.

The archived launch statement said the founders had started "the world's first straight-shot SSI lab, with one goal and one product: a safe superintelligence." It called SSI "our sole focus."

SSI calls itself "an American company with offices in Palo Alto and Tel Aviv." According to TechCrunch, Gross previously served as a Y Combinator partner, while Levy worked as an OpenAI engineer. SSI says it plans to "advance capabilities as fast as possible while making sure our safety always remains ahead."

According to CNBC, citing unnamed sources, Meta tried earlier in 2025 to acquire SSI and to hire Sutskever, and Sutskever rebuffed both efforts. Meta then moved to hire Gross. For Meta's own lab, see Meta Superintelligence Labs explained.

On July 3, 2025, an SSI update note confirmed Gross departed SSI on June 29, 2025, naming Levy President and formalizing Sutskever as CEO. The note also addressed "rumors of companies looking to acquire us," saying SSI was "focused on seeing our work through." The guide Safe Superintelligence explained covers SSI's structure and funding in more detail.

SSI Capitalization and NVIDIA Partnership

SSI has confirmed one $1 billion raise, and press reports put its total backing higher.

On September 4, 2024, SSI announced raising $1 billion from investors including NFDG, Andreessen Horowitz (a16z), Sequoia Capital, DST Global, and SV Angel. A report from TechCrunch noted the raise exceeded $1 billion, citing Reuters regarding a reported $5 billion valuation that SSI did not confirm.

On April 12, 2025, TechCrunch, citing the Financial Times, reported an additional $2 billion funding round at a $32 billion valuation led by Greenoaks. SSI did not comment, and TechCrunch wrote that its product was "presumably" still in development.

On July 27, 2026, SSI and NVIDIA announced a long-term strategic partnership, per a GlobeNewswire release. The release says NVIDIA "has additionally made an investment in SSI" without disclosing the amount. Access to NVIDIA's Vera Rubin platform will let SSI "increase its compute by an order of magnitude."

Reuters, in a story syndicated by The Star, reported a $5 billion NVIDIA equity investment, citing one person briefed on the deal. Reuters said Bloomberg reported the amount first. In the GlobeNewswire release, Sutskever said: "We have research that is worthy of scaling up, and having access to a big NVIDIA computer will let us do so."

Academic Training and Neural Network Research

Ilya Sutskever studied at the University of Toronto, where Geoffrey Hinton supervised his PhD.

Sutskever gained admission to the mathematics program after Grade 11, earning an Honours Bachelor of Science in Mathematics in 2005, per the University of Toronto. According to Wikipedia, Sutskever holds citizenship in Israel and Canada.

Sutskever pursued graduate research in computer science under Hinton, earning a PhD in 2013. His doctoral dissertation, titled "Training Recurrent Neural Networks," was issued on August 13, 2013, through the University of Toronto repository.

In 2012, Sutskever co-authored the AlexNet paper, "ImageNet Classification with Deep Convolutional Neural Networks," with Alex Krizhevsky and Geoffrey Hinton. The paper, published for NIPS 2012, reported a top-5 test error rate of 18.9%.

The University of Toronto noted that the AlexNet source code will be preserved at the Computer History Museum.

Google Brain and Sequence-to-Sequence Architecture

In 2012, Sutskever, Hinton and Krizhevsky founded DNNresearch Inc.

In March 2013, Google acquired DNNresearch, as announced by the University of Toronto. At Google Brain, Sutskever worked as a research scientist, and U of T says he contributed to AlphaGo.

During his tenure at Google Brain, Sutskever co-developed the sequence-to-sequence learning framework alongside Oriol Vinyals and Quoc V. Le. Their paper, "Sequence to Sequence Learning with Neural Networks," was first submitted to arXiv on September 10, 2014.

In November 2024, the paper received the NeurIPS 2024 Test of Time Paper Award, shared that year with the generative adversarial network (GAN) paper.

In 2015, MIT Technology Review selected Sutskever for its annual Innovators Under 35 list in the Visionaries category while at Google Brain.

OpenAI Leadership and the Superalignment Team

Ilya Sutskever co-founded OpenAI in December 2015, serving as research director and later chief scientist.

The founding announcement, titled Introducing OpenAI, named Sutskever as OpenAI's research director. NVIDIA's release later credited him with contributions to GPT models and research leading to reasoning models.

In July 2023, OpenAI named Sutskever and Jan Leike co-leads of its new Superalignment team. The official announcement on the OpenAI blog set a four-year goal and dedicated "20% of the compute we've secured to date" to the effort.

According to reporting from Fortune, citing six anonymous sources, the team never received the pledged 20% compute allocation, though one source disputed parts of that account.

Following the resignations of Sutskever and Leike in May 2024, OpenAI disbanded the Superalignment team. OpenAI announced Jakub Pachocki as Chief Scientist on May 14, 2024, as recorded in the departure statement. For how researchers define superintelligence, see the guide to what superintelligence is.

Board Transition and Departure from OpenAI

Sutskever held an OpenAI board seat during the November 2023 leadership change.

On November 17, 2023, the OpenAI board announced that Sam Altman would depart, per the official leadership statement. The announcement stated Altman was not consistently candid in his communications with the board, which included Sutskever, Adam D'Angelo, Tasha McCauley, and Helen Toner.

On November 20, 2023, Sutskever posted on X that he deeply regretted his participation in the board's actions, as reported by TechStartups. Sutskever wrote that he never intended to harm OpenAI. He said he would "do everything I can to reunite the company."

Altman returned as chief executive on November 29, 2023, accompanied by a reconstituted initial board. In the return announcement, Altman noted Sutskever would no longer serve on the board while expressing hope to maintain their working relationship.

OpenAI announced on May 14, 2024 that Sutskever would leave, and Altman wrote that Sutskever "has something personally meaningful he is going to go work on."

Views of Ilya Sutskever on Pre-Training Limits and Scaling

Sutskever used his NeurIPS 2024 award talk to argue that pre-training will run short of data.

While accepting the Test of Time award at NeurIPS 2024 in Vancouver on December 13, 2024, Sutskever discussed data limits, as reported by MarketScreener. He said "pre-training as we know it will unquestionably end."

According to coverage from Cointelegraph, Sutskever said "Data is not growing because we have but one internet." He called data "the fossil fuel of AI" and said "we've achieved peak data, and there will be no more."

Professional Honours and Recognition

In 2022, the Royal Society elected Sutskever as a Fellow. He was named to the TIME100 AI 2023 list as OpenAI's co-founder and chief scientist, and again to TIME100 AI 2024.

On June 6, 2025, the University of Toronto awarded Sutskever an honorary Doctor of Science.

In 2026, the National Academy of Sciences gave Sutskever its Award for the Industrial Application of Science. NAS says this was the first time the award went to work in AI.

To compare Sutskever with other AI leaders, see AI people to watch.