Jeff Bezos's "Project Prometheus" is raising $10B at a $38B valuation to build "Physical AI".
Jeff Bezos's five-month-old AI startup, Project Prometheus, is closing a $10 billion funding round at a $38 billion valuation, backed by JPMorgan and BlackRock. The company is building AI that natively understands physics to redesign how aerospace, automotive, and robotics produc...
According to reporting from the Financial Times, published on April 21, 2026, Jeff Bezos's secretive physical AI laboratory is nearing the close of one of the largest single funding rounds in the history of artificial intelligence. The company, known as Project Prometheus, was founded roughly five months before this announcement, placing its birth around November or December 2025. With JPMorgan and BlackRock anchoring the round, this is not a venture capital bet. This is Wall Street writing a very large check on a very specific vision.
Why This Matters
A five-month-old company raising $10 billion is not normal, and calling it aggressive would be an understatement. The $38 billion post-money valuation puts Project Prometheus ahead of where most AI companies land at their public debuts, which tells you exactly how much weight Bezos's name and the physical AI thesis carry with institutional investors right now. BlackRock manages roughly $10 trillion in assets globally, and JPMorgan is one of the most conservative capital deployers on the planet. When those two firms anchor a round like this, the rest of the market pays attention. This is the biggest signal yet that "physical AI," meaning AI trained to reason about the real, physical world rather than just text and images, is about to become a serious competitive battleground.
Daily briefing from 50+ sources. Free, 5-minute read.
The Full Story
Project Prometheus emerged quietly from Jeff Bezos's post-Amazon chapter as a focused bet on a specific and underserved corner of artificial intelligence. While most of the AI industry has spent the last three years racing to build better chatbots and coding assistants, Bezos appears to have concluded that the real value sits somewhere else entirely: in the physical world, where aircraft need to be lighter, car batteries need to last longer, and robots need to move through unpredictable environments without falling over.
The company describes its core technology as "Physical AI," a term meant to describe AI systems that do not merely process data but understand the underlying laws of physics well enough to apply them to real engineering problems. That is a genuinely different ambition than what OpenAI, Anthropic, or Google DeepMind are primarily chasing. The hypothesis is that if you give an AI system a deep, native grasp of physics, it can help engineers design better products faster and at lower cost than any human team working alone.
The three industries Project Prometheus has targeted, aerospace, automotive, and robotics, were not chosen randomly. All three involve systems where physics is not a background consideration but the central design constraint. An aircraft's aerodynamics, an electric vehicle's battery chemistry, a robot's grip force calculations: these are domains where even small improvements in physical modeling translate into enormous cost savings and performance gains. The commercial logic is hard to argue with.
The funding structure itself is worth examining closely. The $10 billion round brings total capital raised to more than $16 billion in five months, which suggests there was meaningful early funding before this anchor round closed. The post-money valuation of $38 billion means the pre-money figure sat around $28 billion, giving early investors and Bezos himself considerable paper value before JPMorgan and BlackRock wrote their checks. Additionally, reporting references a holding company structure pitched at a $100 billion valuation, hinting that Prometheus may be just one piece of a larger corporate architecture Bezos is assembling.
The choice of institutional investors over traditional venture capital is also deliberate and meaningful. Venture firms take big swings on unproven technology. JPMorgan and BlackRock do not. Their involvement signals that Project Prometheus has already made a credible case for near-term commercial viability, not just long-term scientific promise. That is a different conversation entirely from the ones happening at most AI seed rounds.
Key Details
- Project Prometheus was founded approximately in November or December 2025, making it roughly five months old at the time of the April 21, 2026 announcement.
- The funding round totals $10 billion at a $38 billion post-money valuation.
- Total capital raised since founding now exceeds $16 billion.
- Anchor investors include JPMorgan Chase and BlackRock, the world's largest asset manager with approximately $10 trillion in assets under management.
- Target industries include aerospace, automotive, and robotics.
- A broader holding company has been pitched at a $100 billion valuation, according to sourced reporting.
- The pre-money valuation implied by the round structure is approximately $28 billion.
What's Next
Watch for Project Prometheus to announce its first major industry partnerships in aerospace or automotive within the next six to twelve months, since institutional investors at this scale expect to see commercial traction quickly. The rumored $100 billion holding company structure suggests Bezos may be preparing to roll up adjacent physical AI and robotics companies under a single umbrella, so acquisition activity should be on everyone's radar. If the technology performs anywhere near its stated ambition, this company will be in a direct collision course with existing engineering software giants like Ansys and Siemens Digital Industries, both of which have built billion-dollar businesses on physics simulation.
How This Compares
The closest parallel in recent memory is the rise of xAI, Elon Musk's AI company, which raised $6 billion in May 2024 and then closed a $12 billion round in December 2024, reaching a $50 billion valuation. Like Project Prometheus, xAI benefited enormously from a famous founder's credibility and moved at a speed that traditional startup timelines cannot explain. The difference is that xAI is competing directly with OpenAI in the large language model space, which is already crowded. Bezos has chosen a lane with fewer direct competitors and a clearer path to industrial contracts.
Compare this also to Figure AI, the humanoid robotics company that raised $675 million in February 2024 at a $2.6 billion valuation, with backing from Microsoft, OpenAI, Nvidia, and Jeff Bezos himself. Figure AI is targeting physical robotics from the hardware side. Project Prometheus appears to be attacking the same problem from the software and AI reasoning side, which means these two companies could end up as natural partners or, if their ambitions expand, natural rivals. The fact that Bezos invested in Figure AI before founding Prometheus is worth tracking.
You should also put this next to what Nvidia has been building with its Isaac platform for robotics and its Omniverse simulation environment, both of which are physics-aware AI tools aimed at similar industrial applications. Nvidia has the hardware advantage and an existing developer ecosystem. Project Prometheus is betting it can build a more sophisticated physics understanding layer on top of whatever hardware stack wins. That is a plausible bet, but Nvidia is not a company that cedes ground easily. The competition in physical AI tools and platforms is about to get significantly more interesting, and you can read more about the broader developments shaping this space in our AI news coverage.
FAQ
Q: What is Physical AI and how is it different from regular AI? A: Physical AI refers to artificial intelligence systems designed to understand and apply the laws of physics, not just process text or images. Regular AI, like a chatbot, learns patterns in language. Physical AI learns how objects move, deform, and interact in the real world, which makes it useful for engineering products like aircraft, cars, and robots rather than answering questions.
Q: Why is a five-month-old company worth $38 billion? A: The valuation reflects investor confidence in Jeff Bezos's track record, the size of the industrial markets being targeted, and the scarcity of credible competitors in physics-native AI. Aerospace, automotive, and robotics together represent trillions of dollars in annual economic activity, and even a small efficiency gain in those sectors justifies enormous upfront investment from institutions like BlackRock and JPMorgan.
Q: Who is funding Project Prometheus and why does that matter? A: The anchor investors are JPMorgan and BlackRock. BlackRock manages roughly $10 trillion in assets globally as of 2026, and JPMorgan is one of the largest banks in the world. Unlike venture capital firms that routinely fund speculative bets, these institutions typically require strong evidence of commercial viability before committing capital at this scale, which makes their involvement a meaningful signal of credibility.
Project Prometheus is the kind of company that changes how an entire category of investment gets perceived, and Bezos has built enough credibility to attract the capital needed to find out whether the physics-AI thesis is correct. If it works, the implications for how physical products are designed and built over the next decade will be substantial. Subscribe to the AI Agents Daily weekly newsletter for daily updates on AI agents, tools, and automation.
Get stories like this daily
Free briefing. Curated from 50+ sources. 5-minute read every morning.
