Jeff Bezos is back. The Amazon founder has cut his retirement short to spearhead his new startup, Prometheus, a new AI venture that is instantly forcing its way into the same conversation as OpenAI and Anthropic. But Bezos isn’t here to replicate what’s already been done, not another AI chatbot or vibe coding platform.
Backed by a massive new funding round that places the company among the elite of Silicon Valley before even shipping a product, Prometheus is focusing on an entirely new frontier – AI in physical engineering and heavy industry.
What is Prometheus all about
For a company to be named after the Greek God of forethinking, Bezos had a vision to use AI where it could bring a revolution in engineering. Prometheus’ goal is to utilise the cutting-edge neural networks and latest artificial intelligence advancements in the manufacturing sector. It wants to hand over AI to engineers for building clever things faster and smarter. Unlike its namesake, it wants to enable humans to build the next fire.
For all this time, we had a basic idea as to what Prometheus wanted to do. Its founder has now secured more funding in the latest round, thereby making it one of the more valuable AI companies that is yet to ship a product.
Even though his other company is in the business of AI chatbots and platforms, Bezos’ new venture is getting into an area where no major players exist at the moment – no chatbots, no vibe-coding assistants, no hardware business.
After stepping down as Amazon CEO in July 2021, Bezos went into retirement but eventually cut it short. In November 2025, he joined hands with Google veteran Vik Bajaj to establish Prometheus as a new venture. The company is all over the news after it closed a staggering $12 billion Series B funding round, thus catapulting its valuation to $41 billion!
Since it’s Bezos we are talking about, the funding round was backed by the cream of Wall Street names, including JPMorgan Chase, Goldman Sachs, BlackRock, DST Global, and Arch Venture Partners. It also saw a heavy personal investment from Bezos himself, bringing Prometheus’s total funding to more than $18 billion.
That’s a lot of funding for a company that wants to put AI in the hands of engineers. But the money is only half the story.
Bringing AI to the heavy industry
For the past few years, generative AI has largely remained accessible via computer screens, automating text, images, and software code. Prometheus wants to do “Physical AI” – the application of neural networks to the complex and high-consequence world of tangible manufacturing.
Prometheus will be targeting industries such as aerospace, automotive, advanced manufacturing, data center infrastructure, and drug discovery, wherein the company wants to move the iterative trial-and-error and, most importantly, expensive, process of physical engineering entirely into the digital realm. The startup’s goal is to give engineers tools to model, simulate, test, and manufacture physical assets, ranging from microchips to jet engines and bridges. For manufacturers, this is expected to cut down costs and time drastically, thus encouraging innovation and development.
However, utilising AI for the physical world presents a unique roadblock – data scarcity. Unlike large language models (LLMs), which can be trained by scraping billions of words from the internet, there is no online repository containing the precision blueprints, metallurgical stresses, or factory-floor experimental metrics that human engineers require to build hardware safely.
But Bezos’s company knows how to solve it. The company is training its AI models on physics-bound workflows, robotics interactions, and empirical laboratory tests rather than data from the internet. It is training its AI out in the real world, making them learn how things are designed, manufactured and put into production.
If industry rumours are to be believed, Prometheus has ideas to simply buy the factories that generate data for its AI models, all so as to train its AI models better. It is said that the company wants to raise an additional $100 billion holding fund for the same.
Will Prometheus give rise to ‘Artificial General Engineer’?
Yes. In fact, Prometheus’ co-founders are coining the term Artificial General Engineer (AGE).
Unlike Artificial General Intelligence (AGI), which refers to a broad and human-like cognitive capacity across all general tasks (hence ‘general’ in the name), an Artificial General Engineer narrows its focus to the hard sciences. It aims to learn and master the laws of physics, chemistry, and fluid dynamics – all in a bid to automate the end-to-end “design-to-manufacturing” loop.
“The idea that you might build a set of tools that could actually do engineering — an artificial general engineer — is a dream that people have thought about for decades, but it’s never really been possible,” Bezos said in a recent interview. “But now it is, and that’s what we’ve been working on.”
With end-to-end AI software in engineering, Prometheus claims it can compress the standard research and development cycle of physical products tenfold or more. Technically, an aerospace engineer using Prometheus’s platform wouldn’t just use Computer-Aided Design (CAD) to visualise a wing – the AI system would actively generate, test, stress-simulate, and optimise the component’s blueprint for factory production lines autonomously.
Will this replace human engineers then? Bezos frames the mission not as a tool for labour replacement, but as an exponential catalyst for human innovation. “All societal wealth is driven by invention,” Bezos noted, drawing a historical line from the plough to the steam engine. “What Prometheus seeks to do is to offer a set of tools that dramatically accelerates that invention loop.”
Building an AGE requires a lot of capital. Prometheus’s work, just like AI companies, is compute-intensive and hence needs real-world simulation data to train the models. While independent infrastructure providers are helping the company’s AI models to learn, the startup will eventually need hyperscaler brands like Amazon’s AWS to store and process all that data for real-world applications.
Despite the recent funding rounds, Prometheus remains a surprisingly lean operation, with only 150 people working across strategic hubs in San Francisco, London, and Zurich. The company has quietly poached elite researchers and engineers from the core of the AI talent market, including OpenAI, Google DeepMind, Meta, and Nvidia.
Will Prometheus redefine the manufacturing industry as we know it? Only time will tell, but for now, we can say Bezos’ newest firm has the potential.