Jeff Bezos’s physical AI startup Prometheus has raised $12 billion in new funding, giving the company a $41 billion valuation and placing one of the largest bets yet on artificial intelligence built for engineering, manufacturing, and real-world industrial design.

The company was co-founded by Bezos and Vik Bajaj, a former co-founder of Verily, Google’s life sciences unit. Prometheus is building what it calls an “artificial general engineer,” an AI system designed to help create, test, and manufacture complex physical products faster than traditional engineering teams can.

The new round includes backing from Bezos himself, as well as major financial institutions such as JPMorgan Chase, Goldman Sachs, and BlackRock. The size of the financing immediately puts Prometheus among the most valuable AI startups in the world, even though the company is still young and has not yet fully revealed the details of its technology.

The raise also shows how investor attention is moving beyond chatbots, coding tools, and office productivity software. The next major AI opportunity may be in the physical economy: aerospace, manufacturing, medical devices, robotics, electronics, industrial systems, and drug design.

Prometheus Is Not Another Chatbot Company

Prometheus is different from many of the AI companies that have dominated headlines over the past few years.

OpenAI, Anthropic, Google, Meta, and others have built large models that can write, code, summarize, reason, generate images, and power assistants. Prometheus is aiming at something more industrial. It wants AI to help design and manufacture things that must work in the physical world.

That distinction matters because engineering physical products is much harder than generating digital outputs. A chatbot can be wrong and corrected. A physical product has to meet safety, durability, cost, materials, energy, and regulatory requirements. A jet engine, medical device, drug compound, spacecraft component, or advanced consumer product cannot simply look convincing. It has to function under real constraints.

Prometheus’s ambition is to compress that long design cycle. Bezos has described the process of moving from an idea to manufacturing at scale as slow and difficult. Prometheus is being built around the idea that AI can shorten that cycle by helping engineers explore designs, simulate outcomes, test options, and eventually move products into production faster.

The “Artificial General Engineer” Pitch Is Ambitious

The phrase “artificial general engineer” is intentionally broad. It suggests a system that does more than answer technical questions. Prometheus wants to build AI that can operate across engineering disciplines and help with the full process of making physical things.

That could include product design, materials selection, simulation, testing, manufacturing planning, optimization, and quality control. In theory, such a system could help engineers move from concept to prototype to production with fewer delays.

This is a much harder problem than building a general-purpose chatbot. Physical engineering requires knowledge of mechanics, chemistry, biology, physics, electrical systems, manufacturing methods, supply chains, and safety rules. It also requires models that understand trade-offs. A part may be strong but too heavy. A design may be efficient but too expensive. A material may perform well but fail regulatory checks or be difficult to source.

The promise of Prometheus is that AI could help manage those trade-offs at a speed that human teams cannot match alone. The risk is that the technology may take years to prove at the level required for critical products.

Physical AI Is Becoming a Major Investment Theme

The Prometheus funding round lands during a surge of interest in physical AI.

Investors have become increasingly focused on AI systems that connect to the real world through robots, factories, labs, vehicles, sensors, manufacturing lines, and industrial workflows. The appeal is clear. Software AI can be powerful, but physical AI may be more defensible because it is harder to copy.

A chatbot startup can be challenged quickly if a larger model provider releases a better feature. A company that builds deep industrial systems, proprietary manufacturing workflows, robotics networks, lab automation, or engineering infrastructure may have stronger moats.

That is one reason physical AI has become attractive to venture capitalists and strategic investors. The physical economy is enormous, and many of its workflows remain slow, fragmented, and expensive. If AI can make engineering and manufacturing faster, the economic impact could be huge.

Prometheus is now one of the most visible examples of that thesis. Its valuation suggests investors believe the company could become a foundational platform for industrial AI, not just another specialized tool.

Bezos Gives the Company Unusual Credibility

Bezos’s involvement gives Prometheus immediate weight.

He built Amazon into one of the world’s largest companies, invested heavily in cloud computing through AWS, and founded Blue Origin to pursue long-term aerospace ambitions. That combination of commerce, infrastructure, logistics, compute, and space experience makes him a natural figure for a company focused on physical AI.

Prometheus also benefits from Bezos’s ability to attract capital, talent, and attention. A $12 billion round for a young startup would be extraordinary under almost any circumstances. With Bezos attached, investors appear willing to fund a much larger and longer-term vision.

That does not guarantee success. Building AI for physical engineering is not the same as building e-commerce, cloud services, or rockets. But Bezos has a history of backing expensive, infrastructure-heavy projects that require patience and scale.

The company’s co-founder Vik Bajaj also strengthens the story. His background in life sciences and advanced technology suggests Prometheus is not only targeting mechanical engineering, but also areas such as drug design and medical technology where AI could speed up research and development.

The Funding Shows How Expensive Frontier AI Has Become

The size of the raise also reflects the capital demands of next-generation AI.

Building a serious physical AI platform will likely require massive compute, specialized models, high-quality data, simulation systems, engineering talent, partnerships with manufacturers, and domain-specific validation. Unlike a simple software product, Prometheus may need to work across laboratories, factories, design environments, and industrial supply chains.

That requires more money than ordinary startup development. If Prometheus is training large models for engineering and physical simulation, its infrastructure costs could be enormous. It may also need to collect or license technical data that is difficult to obtain.

This is part of a wider trend. The leading AI companies are raising huge sums because frontier model development is expensive. Compute costs, chip supply, data centers, energy, and technical talent are now central to the AI race.

Prometheus’s funding round suggests investors believe physical AI will require similarly large capital commitments. The difference is that the payoff, if it works, could extend deep into the industrial economy.

The Company Still Has Much to Prove

Despite the huge valuation, Prometheus remains a company with many unanswered questions.

It has not fully disclosed how its technology works, what models it has built, which industries it will target first, or when customers will be able to use its products at scale. It is also unclear how much of the system is currently functional versus still in research and development.

That matters because the “artificial general engineer” concept is difficult to evaluate from the outside. The phrase is powerful, but the real test will be whether Prometheus can produce measurable improvements in real engineering workflows.

Can it design better components? Can it shorten product development timelines? Can it reduce testing costs? Can it help manufacturers move from prototype to production faster? Can it meet the regulatory requirements of industries such as aerospace, medicine, and energy?

Those are hard benchmarks. A model that performs well in a demo may still fall short in real manufacturing. Physical products require reliability, traceability, safety review, and accountability. Engineers and regulators will not accept vague AI outputs for critical systems.

Engineering AI Could Change the Role of Human Experts

Prometheus also raises a major labor question. If the company succeeds, what happens to engineers?

The likely answer is complicated. AI systems could automate parts of engineering work, especially repetitive design exploration, simulation, optimization, documentation, and testing support. That could make teams faster and reduce the number of people needed for some tasks.

At the same time, physical engineering is not only about generating designs. It requires judgment, accountability, field experience, safety responsibility, regulatory knowledge, and coordination across teams. AI may become a powerful copilot for engineers before it becomes a replacement.

The term “artificial general engineer” suggests a future where AI can take on more responsibility. But in high-stakes industries, human experts are likely to remain central for approval, supervision, safety, and final decision-making.

The real disruption may come from changing productivity expectations. If AI can help one engineering team do the work of a much larger group, companies may rethink how they build products, staff teams, and compete on development speed.

The Physical World Is the Hardest AI Frontier

Prometheus is entering one of the most difficult areas of AI.

Digital AI can move quickly because outputs are easy to generate, test, and revise. The physical world is less forgiving. Materials fail. Machines wear out. Supply chains break. Regulations slow down deployment. Testing takes time. Manufacturing at scale introduces problems that do not appear in simulations.

That is why physical AI may take longer to mature than consumer chatbots or coding assistants. But it may also be more valuable if it works. The industries Prometheus is targeting are large, complex, and expensive. Even a modest improvement in the speed of engineering and manufacturing could create enormous value.

This is the core bet behind the $41 billion valuation. Investors are not only betting on another AI interface. They are betting that AI can reshape the way real things are invented and produced.

A Huge Bet on Industrial AI

Prometheus’s $12 billion raise is a signal that the AI race is entering a new stage.

The first stage was dominated by models that could talk, write, code, and generate digital media. The next stage may be shaped by models that can design products, accelerate research, guide manufacturing, and operate inside the physical economy.

Bezos and Bajaj are positioning Prometheus at the center of that shift. The company wants to build AI that can understand engineering deeply enough to help make complex products faster. If that vision works, Prometheus could become one of the most important AI companies of the decade.

But the distance between vision and proof is large. The company still has to show that its models can handle real industrial complexity, satisfy customers, and deliver results in fields where mistakes are expensive.

For now, Prometheus is one of the clearest signs that investors believe AI’s biggest opportunity may not be another app. It may be the redesign of how the physical world gets built.