Sources said
Kenrick Cai, Krystal Hu and Anna Tong
(Reuters) – Safe Upertelligence, an artificial intelligence startup co-founded last year by former Openai chief scientist Ilya Sutskever, is negotiating to raise funds at a valuation of at least $20 billion.
That would quadruple the company’s $5 billion valuation in its last round of funding in September, when it came from including Sequoia Capital, Andreessen Horowitz and DST Global Five investors raised $1 billion.
SSI's fundraising tests tested industry-wide reassessment by high-profile AI companies, and continued premiums after Chinese startup DeepSeek unveiled its low-cost AI last month.
SSI, which has not generated any revenue, said its mission is to develop “secure superintelligence” that is smarter than humans while being consistent with human interests.
Sources said this week that the company's conversations with existing and new investors are still in an early stages and terms may still change. It is not clear how much money SSI is trying to raise.
Founded in June, SSI has offices in Palo Alto and Tel Aviv and has no response to requests for comment. Sutskever's co-founder is Daniel Gross, who previously led the AI initiative at Apple, while former OpenAI researcher Daniel Levy is Daniel Levy .
Secret startup company
Apart from a rough explanation of the company’s goals for security AI, little is known about secret startups or their work. The interest aroused by investors is Saskoff's reputation and the novel approach he says the team is working on.
In the AI circle, he is a legend of his contribution to breakthroughs because of his investment fanaticism in generating AI. He was an early advocate for scaling, which meant that a lot of computing power and data were dedicated to refining AI models.
This concept is the basis for AI, such as Openai’s Chatgpt, setting a wave of waves for bargaining chips, data centers and energy investments.
As the available databases are reduced to the training model, Sutskever also saw potential upper limits for this approach very early on. Recognizing the importance of investing resources in the inference phase, or in the AI phase when a trained model draws conclusions, he formed a team that will become the latest series of OpenAI’s inference models, right after that.