Prometheus builds AI software for designing and manufacturing complex physical products. The startup has now raised $12 billion at a $41 billion valuation from Jeff Bezos, JPMorgan Chase, Goldman Sachs, BlackRock, and other backers. It’s chasing a brutal bottleneck in industry: turning an idea into a manufacturable machine, device, or compound still takes years, giant teams, and a lot of expensive iteration. Bezos and Vik Bajaj launched the company in November 2025, betting that AI can compress that cycle in a way older engineering tools never could.
What does Prometheus physical AI actually build?
Prometheus is building what Bezos calls an “artificial general engineer” — software meant to help companies design physical systems and move them from concept toward manufacturable reality. Bezos has described it as a “very, very modern version” of CAD, but that undersells the ambition. The idea isn’t just drawing parts on a screen. It uses AI across design and prototyping. Testing, calibration, and pre-production work are part of it too.
That distinction matters. Bezos has been explicit that Prometheus isn’t a robotics company and isn’t mainly about automating factory floors. He’s framed it instead as software for the messy stage before mass production — where engineering teams cycle through requirements, models, experiments, and revisions until something finally works at scale. Axios described the focus as optimizing pre-production machinery and processes, including prototyping.
The pitch is that a customer could shorten the “dream-build” loop. A jet engine maker trying to squeeze out 10% more thrust, for example, might not need to grind through a decade-long program just to explore the design space. The same logic could apply beyond aerospace. Think medical devices, electronics, or even drug compounds. Bezos has already pointed to Blue Origin as the kind of customer that could benefit, though he says Prometheus stands on its own rather than sitting inside Blue or Amazon.
Who founded Prometheus physical AI and why now?
The founding story
Prometheus emerged from stealth in November 2025 with an eye-watering $6.2 billion initial raise, then followed that with this much larger Series B in June 2026. Bezos said he got so excited about the idea that he became co-CEO — his first operational role at a tech company since stepping down as Amazon CEO in 2021. The company’s core thesis is simple to say and very hard to execute: the pace of physical creation has lagged way behind the pace of software, and AI might finally close that gap.
Why Bezos and Bajaj make sense together
Bezos brings obvious industrial scar tissue. He built Amazon into a giant logistics and computing machine, still serves as executive chairman, and has spent years pushing Blue Origin through the painfully slow realities of designing hardware that can’t fail. That matters here. Prometheus isn’t trying to automate spreadsheets. It’s going after engineering work where errors are expensive and timelines can stretch into the next decade.
Bajaj is less famous, but he’s arguably the more natural fit for the technical thesis. He co-founded Google Life Sciences, later renamed Verily, and served as its chief scientific officer. After that, he became chief scientific officer at Grail before moving into Foresite Capital and then helping build Foresite Labs. That mix — hard science, applied research, and company-building — is the sort of résumé you’d want for a startup trying to merge AI with engineering workflows.
Early signals from the company
Prometheus is still unusually secretive for a startup carrying this much capital. What’s public is enough to show scale, though: the company has about 150 employees and offices in San Francisco, London, and Zurich. Bezos has also said a large share of the money will go toward compute. That tells you this isn’t a lightweight SaaS play. It’s a capital-intensive R&D effort that expects very big models and very costly experimentation.
The financing behind the bet
This new round values Prometheus at $41 billion and includes money from Bezos, JPMorgan Chase, Goldman Sachs, BlackRock, DST Global, Arch Venture Partners, and others. It follows the $6.2 billion launch financing from late 2025 and is widely described as a Series B. That’s an absurd amount of money for a company still keeping most product details hidden. Investors aren’t backing traction here so much as a theory of where engineering software goes next.
Where Prometheus sits against competitors
Prometheus overlaps with physical AI, but not in the same way as robotics-first labs. Physical Intelligence has raised $600 million at a $5.6 billion valuation to build AI software for robots, while Skild AI raised $1.4 billion at a $14 billion valuation around its general-purpose robot intelligence system. Prometheus is aiming earlier in the chain. Its real fight is with the fragmented stack of CAD and simulation tools. Prototyping and engineering review still depend heavily on human iteration and institutional memory.
Why are investors backing Prometheus physical AI?
Because this kind of company can’t be built on a normal startup budget.
A product that touches design and testing, then reaches into manufacturing, has to earn trust in ways a chatbot never does. It needs compute, domain specialists and time. Bezos has already said compute is a major use for the new capital, and that lines up with the broader thesis: if Prometheus works, it could become core infrastructure for industries where one better engineering decision can save years and billions.
There’s also a labor argument behind the round, and it’s a controversial one. Bezos has pushed back on the idea that AI automatically means mass white-collar unemployment, saying instead that AI-driven productivity could create “labor scarcity.” He put it bluntly: “Significant productivity in the economy is going to raise the standard of living.” You can disagree with that. A lot of people do. But it helps explain why Prometheus isn’t pitching itself as a cost-cutting side tool. It’s pitching itself as a way to let smaller teams take on much bigger engineering problems.
How big is the market for AI in manufacturing and engineering?
The money chasing this category isn’t random. Grand View Research forecasts the global AI in manufacturing market will reach $47.88 billion by 2030, with a 46.5% compound annual growth rate from 2025 through 2030. That’s the kind of curve investors love, especially when the buyers are large industrial companies that can spend heavily once the software proves itself.
The timing makes sense too. Manufacturers are trying to connect design, automation, analytics, and production data instead of running them as separate islands. PwC has argued that industrial winners by 2030 will be the companies that can orchestrate AI, automation, analytics, and engineering on shared workflows rather than just buying isolated tools. Prometheus is basically a giant wager that engineering itself is ready for that same integration.
Is Prometheus physical AI the next big industrial AI bet?
Maybe. But it hasn’t earned that label yet.
The valuation is massive. The secrecy is real. The product still sounds more like a thesis than a finished platform. Even so, Prometheus physical AI is trying to solve a problem that matters: the painfully slow path from idea to manufacturable thing. What matters next is whether Bezos and Bajaj can show named customers, real workflow gains, and proof that their software beats the old engineering stack on speed, cost, or both.
Read how 4baseCare raised ₹128 Cr in Series B funding to expand its genomics lab network and scale OncoTwin, an AI-powered oncology platform designed to improve cancer treatment decisions across emerging markets.
FAQ
- What funding did Prometheus announce? Prometheus announced a $12 billion Series B at a $41 billion valuation in June 2026. The round included Jeff Bezos, JPMorgan Chase, Goldman Sachs, BlackRock, DST Global, and Arch Venture Partners, following the company’s earlier $6.2 billion launch financing in November 2025.
- How does Prometheus work? Prometheus is building AI software that helps engineers move from concept through design and prototyping. It also covers testing, calibration, and pre-production planning. Bezos has described it as a next-generation CAD-like system, but the point is broader: cut down the years of manual iteration that slow hardware and industrial development.
- Who founded Prometheus? Prometheus was co-founded by Jeff Bezos and Vik Bajaj. Bajaj previously co-founded Google Life Sciences, later called Verily, served as Grail’s chief scientific officer, and went on to help lead Foresite Labs — a background that gives him credibility in complex science-heavy systems.
- Is Prometheus part of the physical AI market? Yes, but it sits in a specific corner of that market. Prometheus is part of the physical AI push because it targets engineering and manufacturing, yet Bezos has said it’s not a robotics company; the focus is the design and pre-production layer rather than training robots to act in the world.
