Jeff Bezos Closes In on $10 Billion for Project Prometheus, His Physical AI Lab

Jeff Bezos is on the verge of closing a $10 billion funding round for Project Prometheus, his ambitious AI laboratory focused on building systems that understand the physical world. The round values the San Francisco-based startup at $38 billion — making it one of the most valuable AI companies to emerge in 2026 — and follows an initial $6.2 billion raise at the company's November 2025 launch.

Project Prometheus is co-led by Bezos and Vikram "Vik" Bajaj, who serve as co-CEOs, marking Bezos's first hands-on operational role since stepping down as Amazon CEO in 2021. The lab's mission is to develop "physical AI": models trained on real-world experimental data, robotics interactions, and engineering workflows rather than text alone. This positions Prometheus in a distinct lane from language-model-focused labs, targeting use cases across manufacturing, aerospace, robotics, drug discovery, and logistics automation.

JPMorgan and BlackRock are among the institutional investors backing this round — a notable signal of confidence from the traditional finance sector. The company has been aggressively recruiting top talent, most recently bringing on Kyle Kosic, co-founder of xAI, and operates additional offices in London and Zurich alongside its San Francisco headquarters. Once the current round closes, total funding will exceed $16 billion.

Bezos's ambitions extend beyond model development. Reports indicate he is planning a holding company to acquire industrial businesses whose operational data could be fed back into Prometheus's AI models — a vertically integrated strategy that would give the lab a self-reinforcing data advantage. Training on proprietary engineering and physics data from real industrial operations could prove far more valuable than any synthetic dataset.

The announcement arrives as physical AI and robotics attract surging investment across the industry, with competitors including Figure AI and Physical Intelligence also raising massive rounds. Prometheus's physics-first approach — grounding AI directly in the laws governing the real world rather than statistical patterns in text — may represent one of the most direct paths to systems that can reliably operate in the messy complexity of physical environments.