Project Prometheus: Jeff Bezos is taking up CEO's job again, and it's not Amazon

Jeff Bezos, the founder of Amazon and one of the world’s richest men, is committing both his time and money to a new artificial intelligence start-up where he will serve as co-chief executive. The company, named Project Prometheus, is launching with $6.2 billion in funding, partly contributed by Bezos. According to three people familiar with the development, who spoke anonymously as details remain undisclosed, this makes it one of the most heavily financed early-stage start-ups globally, reported The New York Times.

This marks the first time Bezos has taken on a formal operational position since leaving his role as Amazon’s chief executive in July 2021. Although he remains deeply involved with Blue Origin, a competitor to SpaceX, he holds only the title of founder there. Since stepping away from Amazon, he has drawn attention for both his personal life, including a celebrity-filled wedding in Venice this year, and for a renewed focus on Blue Origin as well as growing interest in the race to advance artificial intelligence.

More from Tech

US-China tech war has an unexpected winner: AI prodigy with $23 bn fortune Apple ramps up search for new CEO ahead of Tim Cook's exit next year

Prometheus enters a competitive AI landscape

The new company places Bezos at the centre of the increasingly competitive AI sector, where smaller firms are attempting to define their space alongside major players such as Google, Meta and Microsoft, and specialist companies including OpenAI and Anthropic. Project Prometheus has operated quietly until now, and its founding date remains unclear. The start-up will work on AI designed to assist engineering and manufacturing in areas such as computing, aerospace and automotive technologies, aligning with Bezos’ wider interest in space exploration. Its base of operations has not yet been confirmed.

Bezos will lead the company alongside co-founder Vik Bajaj, a physicist and chemist known for his work with Google co-founder Sergey Brin at Google X, often referred to as “The Moonshot Factory”. The research division produced projects like Wing, a drone delivery service, and the autonomous vehicle initiative that evolved into Waymo. In 2015, Bajaj helped found Verily, a life sciences research lab under Alphabet. He later co-founded Foresite Labs in 2018, leaving that role recently to focus on Project Prometheus, according to the same three people cited by the New York Times.

Project Prometheus is part of a growing wave of companies applying AI to physical tasks such as robotics, drug development and scientific research. This year, several researchers departed Meta, OpenAI, Google DeepMind and other major AI organisations to establish Periodic Labs, which aims to accelerate discoveries in physics and chemistry. Last year, Bezos invested in Physical Intelligence, a company that applies AI to robotics. The substantial $6.2 billion funding pool positions Project Prometheus competitively within the high-cost race to build advanced AI technologies; by comparison, Thinking Machines Lab, formed by former OpenAI employees, secured $2 billion this year.

Quick Reads

View All

China launched ‘highly sophisticated’ AI-led espionage campaign — here’s how it was disrupted

'Is it just a sock?' Why Apple's iPhone pocket has got the internet talking

The start-up has already recruited nearly 100 employees, including researchers from leading AI institutions such as OpenAI, DeepMind and Meta, according to the three people familiar with the hiring efforts.

Major AI organisations — including OpenAI, Google and Meta — are already developing technologies intended to accelerate progress in the physical sciences. Two researchers at Google DeepMind recently won a Nobel Prize in Chemistry for AlphaFold, a system advancing drug discovery. These companies frequently state that large language models, the systems behind tools such as ChatGPT, are approaching the ability to drive breakthroughs in areas including mathematics and theoretical physics.

However, firms like Periodic Labs and Project Prometheus aim to develop AI that learns in more complex ways than language models. Large language models operate by analysing extensive digital text, identifying patterns in sources such as Wikipedia articles and news reports, and mimicking human writing. In contrast, the newer companies seek to build systems that also learn directly from the physical world. Periodic Labs, supported by $300 million, plans to construct a laboratory in Northern California where robots will conduct scientific experiments at scale. Through analysing this physical trial and error, AI could theoretically learn to perform experiments independently.

Project Prometheus is expected to explore similar directions, according to individuals familiar with its plans.