Jeff Bezos's Prometheus Raises $12B to Automate Physical Engineering
Prometheus, the physical AI startup co-founded by Jeff Bezos and Vik Bajaj, has raised $12 billion at a $41 billion valuation. The company launched in late 2025 with an initial $6.2 billion raise. Six months later, the total has doubled.
What It Claims to Be Building
Prometheus calls its target an "artificial general engineer": software capable of automating the design and manufacturing of complex physical systems. The stated examples include jet engines and drug compounds.
That is an ambitious scope. Designing a jet engine and designing a drug compound are not obviously similar problems. Whether a single system can do both well, or whether "artificial general engineer" is load-bearing terminology, remains to be seen.
Who's Backing It
The funding round includes Bezos, JPMorgan Chase, Goldman Sachs, and BlackRock. That is not a typical AI seed round. These are institutions with large infrastructure and manufacturing stakes. They are not betting on demos.
Bajaj's background matters here. He co-founded Verily, Google's life sciences unit, before joining Prometheus. That is relevant experience for the drug compound side of the pitch.
The Compute Allocation
A large portion of the $12 billion will go toward compute. That is increasingly standard language for AI raises at this scale. The physical systems angle likely requires significant simulation workloads, which are compute-intensive even before any inference costs.
Scale Check
At $41 billion, Prometheus is among the most highly valued AI startups ever funded. The company has 150 employees across San Francisco, London, and Zurich. That works out to roughly $273 million per employee in valuation. The ratio is not unusual for early-stage AI companies in 2025-2026, but it is worth noting.
The gap between $41 billion and 150 people will need to close eventually. The next milestone worth watching is whether Prometheus publishes results on either the engineering or life sciences side before its next raise.
Source: Techcrunch