Bezos’s Prometheus Raises $12 Billion to Build an AI That Designs Physical Things
Jeff Bezos just stepped out of stealth mode with one of the most ambitious AI bets of the year. Prometheus, the startup he co-founded with former Google executive Vik Bajaj, raised $12 billion this week at a $41 billion valuation. Backers include JPMorgan Chase, Goldman Sachs, BlackRock, and Bezos himself.
This is the company’s second raise. It launched quietly late last year with $6.2 billion, said almost nothing publicly, and has now come forward with its clearest explanation yet of what it’s actually building.
What an “Artificial General Engineer” Actually Means
Most of the AI industry is focused on artificial general intelligence — software that can reason, write, and converse like a human. Prometheus is chasing something different: an artificial general engineer. Software designed to automate the process of designing and manufacturing complex physical things — jet engines, spacecraft, medical devices, drug compounds.
“We’re building tools that will make it much easier for engineers to design physical objects,” Bezos told CNBC. He compared it to a modern, dramatically more capable version of computer-aided design — the software engineers already use, but supercharged with AI that can reason through entire design and manufacturing cycles, not just draw shapes on a screen.
Bajaj described the ambition as supporting engineers “end to end” — from initial design and prototyping through performance analysis and manufacturing. The goal isn’t to replace human engineers entirely. It’s to compress a process that can take years into something dramatically faster.
Why This Is Different From Most AI Startups
Large language models like Claude and ChatGPT learned by digesting enormous amounts of text from the internet. Prometheus is trying to do something structurally different: train AI on data from the physical world — material properties, manufacturing tolerances, structural stress patterns, chemical interactions — to build systems that understand how physical objects actually behave, not just how language describes them.
“This is an age-old dream,” Bezos said. “The idea that you might build a set of tools that could actually do engineering, an artificial general engineer… it’s never really been possible. But now it is.”
It’s a meaningfully different bet from most of the current AI landscape, which remains concentrated in chatbots, coding assistants, and content generation. Prometheus is part of a broader shift some investors call “physical AI” — applying machine learning to atoms rather than just bits.
The Scale Is Small. The Ambition Is Not.
Despite the eye-watering valuation, Prometheus currently employs around 150 people across offices in San Francisco, London, and Zurich. That’s a tiny team for a $41 billion company — smaller than most mid-sized engineering firms, let alone a business aiming to reshape how physical products get built.
Bezos argues that’s the point. “What we’re doing is building a set of tools that will empower engineers to compress that cycle time and make that dream-build loop be 10 times faster or even more,” he told Axios. The bet is that a small team building the right AI tools can have outsized leverage — making existing engineers radically more productive rather than needing to hire armies of new ones.
An Uncomfortable Tension With Bezos’s Own Track Record
There’s an irony worth sitting with. Bezos remains executive chairman and the largest individual shareholder of Amazon — a company that employs more than 1.5 million people worldwide and has cut tens of thousands of jobs over the past year as CEO Andy Jassy accelerates Amazon’s own automation push.
Bezos has been candid about where he thinks this leads. He told CNBC that the productivity gains AI delivers will create what he calls “labor scarcity” — a world where demand for skilled workers outpaces the available supply, even as automation displaces some roles. Whether that vision plays out as broadly beneficial or sharply disruptive will depend heavily on how fast the technology matures and how the gains get distributed.
It’s Not Alone
Prometheus is the most heavily funded player in physical AI, but it isn’t the only one. Investors are increasingly rotating capital from software toward “hard tech” — chips, robotics, defence technology, and industrial AI — as the market matures and the limits of pure chatbot economics become clearer. Elon Musk, never one to stay quiet about a competitor, called Bezos a “copycat” on X after early reports of the venture leaked last year.
Whether Prometheus or a rival builds the first genuinely useful artificial general engineer remains to be seen. But the size of this week’s raise is a clear signal: serious money believes the next major AI breakthrough might not be a smarter chatbot. It might be AI that can actually build things. For more on how AI is expanding well beyond chatbots and into infrastructure and hardware, see our guide to 10 emerging technologies transforming the future.
Read more tech related articles here.
