IA · 14 June 2026 · 3 min read

Prometheus: Jeff Bezos' Startup Raises $12 Billion for an 'Artificial General Engineer'

In brief: Prometheus, the physical AI startup co-founded by Jeff Bezos and Vik Bajaj, has raised $12 billion in a new funding round, reaching a valuation of $41 billion. The venture seeks to build an 'artificial general engineer' to drastically accelerate industrial design, robotics, and aerospace innovation, redefining real-world automation.

by Team Mocchi's

Prometheus: Jeff Bezos' Startup Raises $12 Billion for an 'Artificial General Engineer'

A Record-Breaking Round for Physical AI

Prometheus, the artificial intelligence startup co-founded by Amazon founder Jeff Bezos and biotechnology expert Vik Bajaj, has announced a massive new capital injection. As reported by Ars Technica, the company raised $12 billion in a funding round backed by financial heavyweights including JPMorgan Chase, Goldman Sachs, and BlackRock, alongside significant personal backing from Bezos himself.

This round follows an initial $6.2 billion funding phase last year, pushing the startup's valuation to $41 billion. Despite its massive capital reserves, the firm operates with a lean team of approximately 150 employees, dedicating the vast majority of its capital to securing compute power and generating high-fidelity physical data through advanced simulation environments.

The 'Artificial General Engineer'

Bezos' vision for Prometheus goes far beyond standard language models or administrative digital assistants. The startup is focused on "physical AI"—applying deep learning principles to physics, materials science, advanced manufacturing, and robotics to build an "Artificial General Engineer" (AGE).

Speaking with The New York Times, Bezos framed the initiative as a historic necessity, arguing that human progress is defined by breakthroughs in physical engineering—from the invention of the plow to the steam engine. Prometheus aims to build tools that dramatically compress this innovation cycle, helping engineers design physical hardware and systems far more rapidly.

Overcoming the Real-World Data Bottleneck

One of the primary roadblocks in physical AI is the scarcity of structured real-world data compared to the vast digital text corpora used to train traditional language models. Vik Bajaj explained to CNBC that designing next-generation technology currently requires thousands of human minds working in tandem with engineering software tools that have remained largely unchanged for decades.

Prometheus' core paradigm focuses on building models capable of understanding and predicting actual physical behaviors. The newly raised capital will fund the infrastructure required to generate proprietary physical training datasets through massive scientific simulations. These datasets will train neural networks to automate complex mechanical and structural testing phases that currently delay manufacturing cycles.

A $100 Billion Ecosystem for the Physical World

Jeff Bezos' ambitions for Prometheus extend well beyond its core research facility. Reports suggest that Bezos and Bajaj are laying the groundwork for a parallel $100 billion investment fund. This fund will target startups positioned to leverage and commercialize the scientific and engineering breakthroughs generated by Prometheus.

The resulting technologies could benefit a wide array of physical industries, ranging from advanced logistics and warehouse automation to superconducting materials and aerospace development at Blue Origin, Bezos' space enterprise. While Prometheus has yet to launch its first commercial products, its unparalleled funding gives it a significant head start in the race to automate physical engineering.

Further reading

All articles on the Mocchi's blog