Jeff Bezos isn’t content to dominate cloud computing, e‑commerce, and private spaceflight. Now, he wants to rewire how artificial intelligence understands the physical world.

According to the Financial Times, Bezos is close to completing a $10 billion funding round for his AI lab Project Prometheus, valuing the company at $38 billion. JPMorgan and BlackRock are among the investors. Once finalized, Prometheus’s total funding will exceed $16 billion, including $6.2 billion raised at launch in November 2025.

That makes Prometheus one of the largest and fastest capital accumulations in AI—especially striking for a company that isn’t building a chatbot.

Beyond Language Models

Prometheus, founded in San Francisco by Jeff Bezos and Vikram “Vik” Bajaj, is aimed squarely at what insiders call physical AI. Instead of training models primarily on text, images, or scraped internet data, Prometheus feeds its systems real‑world experimental data, robotics interactions, and engineering workflows.

Bajaj is key to that ambition. A PhD in physical chemistry from MIT, he cut his teeth at Google X, contributing to early work on Waymo and Wing, before co‑founding Alphabet’s life‑sciences spinoff Verily, and later Xaira Therapeutics, an AI‑driven drug discovery company.

Prometheus’s priorities reflect that background: aerospace, automotive engineering, advanced manufacturing, and pharmaceuticals. The company is betting that the next leap in AI won’t come from better text prediction, but from machines that can reason about atoms, materials, machines—and mistakes in the real world.

Buying Talent and Time

To accelerate that vision, Prometheus acquired General Agents, an agentic‑AI startup co‑founded by former DeepMind researcher Sherjil Ozair. The lab has also recruited aggressively, pulling senior researchers from OpenAI, DeepMind, Meta, and xAI.

This isn’t cheap—and that’s the point. Physical AI requires something most large language models never touch: proprietary, high‑friction data generated through experiments, simulations, and robotic trials. That data takes time, hardware, and capital to produce.

Investors appear confident the payoff will be worth it.

A Crowded but Fragmenting Field

Prometheus enters a sector heating up fast. Periodic Labs has raised $300 million to build robot‑run research facilities. Figure AI is making headway in humanoid robotics, while Tesla and xAI push their own real‑world AI strategies.

Elon Musk has reportedly dismissed Prometheus as a “copycat,” referencing overlaps with xAI’s emphasis on the physical environment. But insiders argue the approach is fundamentally different. Where Musk leans on vertically integrated hardware platforms, Prometheus presents itself as a science‑first lab, with deeper roots in chemistry, materials research, and biomedical engineering.

That distinction matters—especially for industries where error margins are measured in microns, not tokens.

The Bigger Bezos Play

The funding round ties into an even bolder plan. Bezos is said to be exploring the creation of a $100 billion holding company designed to acquire industrial firms disrupted by AI. Their operational data—manufacturing logs, quality‑control records, supply‑chain telemetry—would then feed directly into Prometheus’s models.

It’s a familiar Bezos logic: own the infrastructure, own the data, and let competitors rent access downstream.

Prometheus already has offices in London and Zurich, signaling early expansion into Europe’s industrial core—a region rich in advanced manufacturing expertise but historically cautious about US tech dominance.

A Different Kind of AI Arms Race

At a $38 billion valuation, Prometheus is no longer a skunkworks. It’s a declaration that the next AI war won’t be fought only in chat windows or creative tools—but in factories, labs, and industrial systems that shape the physical economy.

Whether Prometheus delivers remains an open question. Physical AI is harder, slower, and riskier than scaling language models. But if it works, it won’t just answer questions—it will design drugs, build aircraft, and rewire supply chains.

For Bezos, that sounds less like a gamble—and more like unfinished business.

‍