Jeff Bezos Raised 12 Billion Dollars to Build an AI That Designs Jet Engines

Jeff Bezos Prometheus artificial general engineer is the project that pulled the world’s second richest man out of retirement and back into a CEO chair for the first time in five years.

Bezos, who stepped down as Amazon CEO in 2021, now serves as co-CEO of Prometheus alongside Vik Bajaj, a Stanford School of Medicine professor and former co-founder of Google’s life sciences lab Verily. Prometheus announced it raised $12 billion at a $41 billion valuation, with new funding coming from Bezos himself, along with JPMorgan Chase, Goldman Sachs, and BlackRock.

A man who built one of the most valuable companies on earth and then walked away from running it has decided this is the project worth coming back for. That alone is worth paying attention to. What he is actually building is more interesting than the funding number.

What an Artificial General Engineer Actually Is

Prometheus is building what it calls an “artificial general engineer” — software capable of automating the design and manufacturing of complex physical systems, from jet engines to drug compounds.

Think about what that sentence is actually claiming. Not a chatbot that talks about engineering. Not a tool that helps an engineer write reports faster. Software that does the engineering itself the design, the simulation, the manufacturing process the way ChatGPT writes a paragraph.

Bezos described the system as “a very, very modern version of CAD,” computer-aided design software, while cautioning that he is oversimplifying. CAD software has existed for decades. Engineers use it to draw and model parts. What Bezos is describing is a CAD system with judgment built in one that doesn’t just draw what you tell it to draw, but reasons through the physics, the manufacturing constraints, and the trade-offs the way a senior engineer would.

Bajaj framed the engineering challenge plainly: “A jet engine takes teams of engineers a decade just to come up with the design. What has changed is the ability to formulate even something that complicated as an end to end AI problem.”

A decade to design a jet engine. Bezos posed the actual goal directly: “What if instead of a team of 1000 people working for ten plus years to build a next generation of jet engine, what if they could do that in five years or two years or one year? You need to simulate all of the physics and all of the things, even down to the manufacturing processes inside a computer.”

Why This Is Different From Every AI Model You Have Heard Of

This is the detail that separates Prometheus from the ChatGPT and Claude conversation entirely, and it matters for understanding why this is genuinely new rather than just another AI company raising money.

Bezos explained directly: “We have to create our data sets. The training data is completely different from what the LLMs that you’re accustomed to have.”

Large language models like Claude and ChatGPT learned from text scraped across the internet articles, books, code, conversations. There is no equivalent internet scale archive of “how to design a jet engine that doesn’t fail.” Proprietary physics datasets and domain specific training methodologies are the durable advantage Prometheus is building. Bajaj said the data comes partly from established laws of physics and partly from testing results and process data gathered through collaborations with manufacturing companies.

That is a fundamentally different and harder problem than training a chatbot. It means Prometheus cannot simply buy more compute and scale its way to a better model the way language model companies do. It has to build the raw physics knowledge itself, piece by piece, in a domain where mistakes are not embarrassing typos they are jet engines that fail.

The Labor Shortage Argument, and Why It Is Controversial

This is where Bezos says something that sounds backwards at first, and it is worth understanding exactly what he means before judging it.

In a CNBC interview, Bezos pushed back on widespread fears that AI will destroy jobs, predicting instead that productivity gains will cause a labor shortage and even deflation in key sectors.

Bezos said: “I think what’s actually going to happen is we’re going to have labor scarcity as a result. When you have significant productivity in the economy, the standard of living goes up. People who have two earner households might decide one of those earners doesn’t need to be in the job market.”

The historical analogy he leans on is the plow and the steam engine productivity expanded the pie rather than shrinking the workforce.

Here is the part that makes this claim harder to dismiss outright, and also harder to fully trust. Bezos knows something about labor at scale. Amazon, where he serves as executive chairman and remains the largest individual shareholder, employs more than 1.5 million people worldwide, and over the past year, under CEO Andy Jassy, has laid off tens of thousands of workers as the company has accelerated its own automation push.

The man predicting AI will create labor shortages runs a company that is currently laying off workers while automating. That contradiction does not necessarily make him wrong about the long-term economic theory. The plow really did expand farming output without putting most of humanity out of work permanently. But it does mean his prediction is, at minimum, convenient for someone whose other major company is cutting headcount right now.

There is a real counterweight worth naming honestly. NEET levels people not in employment, education, or training have already been rising since 2021, well before the current AI wave hit the workforce. Whether that trend reflects voluntary opt-outs, which is Bezos’s framing, or structural displacement, which is the bear case, is the central question worth watching.

How Big Bezos Thinks This Market Actually Is

The scale of ambition behind Prometheus only makes sense once you see the numbers Bezos and Bajaj are pointing at.

Bajaj said: “We are one small company operating in a physical economy, which amounts to 60% of the world’s GDP. It’s very large, $70 trillion.” For context, US physical sectors alone manufacturing, construction, transportation and warehousing, mining, and agriculture totaled roughly $5.97 trillion, or 19% of US GDP in Q4 2025, with manufacturing alone at just under $3 trillion.

Bezos summarized the philosophy behind the entire bet: “What drives civilizational wealth? And the answer is invention. Somebody invented the plow and we all got wealthier. Somebody invented the steam engine and we all got wealthier. Our goal at Prometheus is building a set of tools that accelerate that invention loop.”

This is not a company trying to capture a slice of an existing market. It is a company betting that compressing the time it takes humanity to invent things is itself the business and that whoever owns that compression owns an enormous share of future economic growth.

The Skeptic’s Case Deserves a Fair Hearing Too

Not everyone is convinced, and the skepticism comes from credible places, not just internet noise.

Elon Musk called Bezos a “copycat” on X after news of the venture leaked last year. The limited details about what Bezos’s AI lab is actually trying to do opens him up to critics who say the billionaire’s endeavor may amount to using his clout to get institutional investors to throw money at chasing AI FOMO.

Coogan, a co-host who covered the raise, noted the round mirrors IPO level numbers, which is why Bezos went directly to JPMorgan, Goldman, and BlackRock rather than traditional venture capital meaning Wall Street is effectively underwriting a private 150 person company at public market size, with public-market consequences if the artificial general engineer thesis falls short of the hardware breakthroughs being priced in.

That is the honest risk in one sentence. A 150 person company, operating mostly in stealth, has convinced some of the largest banks on Wall Street to back it at a $41 billion valuation based largely on Bezos’s track record and a vision still light on public proof. Bezos himself admits it is “premature” to disclose what Prometheus has actually accomplished so far, while insisting “it’s really quite remarkable.”

That is also exactly what someone with nothing to show would say. The difference between visionary patience and an expensive bet that does not pan out is something only time, not this funding round, will actually settle.

BEXORN VERDICT: 7/10 Genuinely Different, Genuinely Unproven

This is not another AI chatbot wrapper dressed up in bigger language. The data problem Prometheus is solving is real and genuinely harder than what OpenAI or Anthropic are doing, and the market it is chasing is enormous and mostly untouched by software so far. That earns real respect.

But a 150 person company with almost nothing public to show, raising money at IPO scale numbers through major banks rather than venture capital, is also exactly the profile of an expensive idea that may never clear the gap between “compelling vision” and “working product.” Bezos has earned the benefit of the doubt through Amazon. He has not yet earned it through Prometheus. Watch for an actual product demonstration before treating this as settled the vision is the easy part, and right now that is all the public has seen.

FAQ

Related Reading

• SoftBank’s Masayoshi Son Says Robots Are the Next Trillion Dollar Bet and He Just Put Real Money Behind It

• SpaceX Bought The AI Coding Tool You Use For 60 Billion Dollars