Jeff Bezos is co-CEO of Prometheus, his first operational role running a company since he left Amazon in 2021. That detail matters. He is not a backer who shows up for board meetings. He told CNBC that Prometheus is the bulk of his time.
The company launched in November 2025 with $6.2 billion and has now raised an additional $12 billion. Total funding exceeds $18 billion, and the valuation sits at $41 billion. For context, the company has 150 employees. That is more capital per person than almost any startup in history.
The core idea is straightforward to describe and difficult to execute. Prometheus wants to do for physical engineering what word-processing software did for writing: make the process faster without replacing the person doing it. Bezos described the product as a very advanced version of the design software engineers already use, but that framing almost certainly undersells the ambition.
The real problem Prometheus is going after is time. A standard jet engine contains close to 25,000 parts, and the design and testing phase alone can take up to five years per model before a single unit is assembled. Making a 10% improvement to an existing engine can require a decade of work. Drug development timelines run five years on average just from clinical trials to approval. These timelines are not a product of laziness. They are a product of complexity, iteration, and the cost of being wrong when a product operates at extreme temperatures or inside a human body.
If Prometheus can compress those cycles by even half, let alone ten times as Bezos claims, the financial impact on any industry that builds complex physical things would be enormous. Faster design cycles mean less capital tied up in prototyping, shorter time to market, and smaller engineering teams needed to achieve the same output.
The investor list tells a story of its own. JPMorgan, Goldman Sachs, and BlackRock are not traditional venture capital firms. They invest at this scale when they see something that could affect the industrial economy as a whole, not just one sector. Their presence suggests this is being read as a structural shift in how physical products get made, not a niche technology play.
There is also a second, larger plan that Prometheus declined to discuss publicly. Bezos has reportedly been in discussions to raise up to $100 billion for a separate holding company that would acquire traditional industrial businesses. The logic is that the hardest problem in building AI for the physical world is training data. Unlike the internet, which produced billions of searchable pages that AI companies could ingest, manufacturing data sits inside private companies on sensors, engineering documents, and production logs. By buying those companies, Prometheus would own that data. It would train the AI on real industrial operations and then sell improved tools back to the companies it already owns.
That structure, if it comes together, would make Prometheus something closer to a Berkshire Hathaway for industrial AI: an owner of real businesses, not just a software vendor.
Bezos's position on jobs deserves a clear reading. He argues that AI will create what he calls "labor scarcity," meaning demand for workers will outpace supply because AI-driven productivity will make the economy grow fast enough to need more people, not fewer. He also said some two-income households might become one-income households by choice as living standards rise. Those two statements point in different directions, and neither resolves the question of what happens to the engineers, technicians, and factory workers whose specific tasks get automated first.
What is clear is that Bezos has built and run operations at Amazon employing over 1.5 million people while simultaneously accelerating automation at every level. He understands both sides of that equation better than almost anyone, which is precisely why his confidence about labor outcomes should be read carefully, not taken at face value.
For business operators in any industry that touches physical manufacturing, this is a company worth watching closely. Prometheus has no public product yet and has shared almost no technical detail. But the money behind it, the people building it, and the problem it is targeting are all real.