Subscribe
AI
3 min read

Inside Prometheus: The Team Jeff Bezos Is Building to Make an Artificial General Engineer

Jeff Bezos has raised roughly $18 billion and hired about 250 people to chase what he calls the artificial general engineer: an AI system that designs and manufactures physical things the way large language models generate text and code. A Fast Company investigation into the workforce of Prometheus, the startup Bezos cofounded and co-leads, found the company is building an end-to-end agentic AI engineering system, and that its hiring pattern says as much about the plan as any press release. Rep

Inside Prometheus: The Team Jeff Bezos Is Building to Make an Artificial General Engineer

Jeff Bezos has raised roughly $18 billion and hired about 250 people to chase what he calls the artificial general engineer: an AI system that designs and manufactures physical things the way large language models generate text and code. A Fast Company investigation into the workforce of Prometheus, the startup Bezos cofounded and co-leads, found the company is building an end-to-end agentic AI engineering system, and that its hiring pattern says as much about the plan as any press release.

Reporter Rebecca Heilweil identified 130 people who say they work at Prometheus, assembled from LinkedIn and personal research pages, out of the roughly 250 total headcount. The company did not recruit the prototypical LLM researcher. It pulled engineers from aerospace, biotech, defense, chemistry, ocean and energy engineering, and high-performance computing, people from xAI, Meta's Reality Labs, Waymo, Epic Games and Red Bull Advanced Technologies. Some came from General Agents, the applied AI lab Prometheus quietly acquired last year, whose former CEO Sherjil Ozair now sits on Prometheus technical staff. An early employee describes the work as accurate, high-fidelity simulations feeding an agentic artificial engineer that could eventually drive simulations and make design decisions at the highest level.

The money and the model

Prometheus launched publicly in November 2025 after being incubated in 2024 at Foresite Labs, the venture studio run by Bezos's cofounder Vik Bajaj, a Stanford professor who previously led science operations at Alphabet's Verily. The company raised an initial round of about $6 billion, Bezos its largest backer, then a $12 billion Series B in June at a $41 billion valuation, with JPMorgan, BlackRock, Goldman Sachs, DST Global and Arch Venture Partners among the investors.

At a Semafor event on November 17, Bezos and Bajaj said most of that capital has gone to compute, and that the company has built one single multi-physics model rather than a suite of narrow simulators. Prometheus has reportedly used neural surrogates for numerical simulation to redesign hardware, including an experiment applying its models to redesign the Mars helicopter. The pitch is compression: a jet engine program that takes a decade of design, prototyping and testing could, in the company's framing, take a few years.

Physical AI targets engineering, not chat

Notably, this is not a robotics play. Bezos has said the company has nothing to do with building robots; he likens the product to a much more advanced version of CAD software, powered by AI. A trademark application mapped ambitions across physical simulation, digital twins, structural, thermal, fluid and acoustic analysis, 3D printing, semiconductor manufacturing and factory layouts.

Why the hiring matters more than the hype

Physical AI needs training data that current models mostly lack: the language of physics, materials and tolerances, not text. That is why a headcount heavy on domain engineers is a technical roadmap disguised as a staffing page. It also signals the customer list. Prometheus retained a technical adviser who serves as chief scientist for the US Army's ground systems research, and an adviser who worked in Marine Corps innovation and at defense contractor Anduril. Bezos has said Blue Origin already uses some Prometheus technology.

The competition is real and young. Anima Anandkumar and Benedikt Jenik were approached about leading Prometheus before starting their own physical-AI startup, Accelerated Understanding; P-1 AI and Neural Concept are building agentic engineering tools from other directions. And the open question, as the skeptics put it, is whether ~250 people and $18 billion of compute are enough to distill engineering itself into one model. Bezos's own framing admits the bet: the dream has existed for decades, and it has never been possible until now.

Sources

AILLMStartups