Jeff Bezos has taken the role of co-CEO in a new AI startup named Project Prometheus. What’s interesting is that this is Bezos’ first position in management since he stepped down as Amazon CEO in July 2021.

Project Prometheus describes itself as a company that focuses on ‘AI for the physical economy.’ Simply put, Prometheus will focus on the application of AI in industries that require advanced engineering and materials science.

This includes all manufacturing activities related to the production of physical goods, such as automobiles, electronics, heavy machinery, and aerospace manufacturing.

Up until now, the entire focus of artificial intelligence has been on ‘bits,’ which is why we have several Large Language Models (LLMs) today powering numerous chatbots and AI assistants.

However, Prometheus wants to focus on ‘AI for atoms,’ where artificial intelligence can be used to stimulate physics, automate factories, design physical systems, and reduce manufacturing times.

Prometheus debuts in an already overcrowded AI market where several small and big players are fighting to create a niche. However, one of Prometheus’ biggest advantages is the huge $6.2B funding it has secured, a large part of which comes from Bezos himself. This makes it one of the most well-funded early-stage startups globally.

In comparison, Periodic Labs, a company that aims to ‘automate scientific discovery’ by focusing on physical AI (much like Prometheus), raised $2B this year. So, of course, Prometheus’ chunky wallet gives it a head-start over its rivals.

China’s Lead in the Physical AI Race

Although the US has attempted to hinder China’s growth in the AI sector through strict export controls on advanced AI-compatible semiconductor chips, the fact remains that China is significantly ahead of the US in terms of physical AI applications.

Data from the International Federation of Robotics shows that China is third on the list of countries with the highest robot density. The country has 470 robots per 10,000 employees, which is up from 402 robots in 2022.

What’s even more commendable is that China entered the top 10 list only in 2019, and in just 6 years, it has raced past countries like Japan and Germany. On the other hand, the US ranks 10th on the list with a density of only 295 robots.

A high robot density points to a highly automated and productive manufacturing sector with relatively fewer defects and faster iteration cycles. Throw a bit of AI tech in there, and you have ‘smart factories’ with automated control systems, AI-backed quality inspections, predictive maintenance, and so on.

Along with this, China also seems to have a large talent pool to tap into:

• China produces 1,059 engineering bachelor’s graduates per million people (2025 estimate) as compared to the US’s 403.
• China graduates 1.5M+ bachelor’s engineers annually, along with 400K master’s and 60K PhDs. By contrast, the US awards just 140K bachelors with 50K master’s and 12K PhDs.

No wonder top US AI startups have to poach talent from other leading companies, something that Prometheus has also done. 

China has already surpassed the US in manufacturing, to the extent that it is now known as the world’s factory. It has left the US behind in areas like advanced battery manufacturing, solar production, EV supply chains, and robotics deployment.

• China accounts for 33% of the world’s electronic exports with $1,368B, whereas the US contributes just 4%. 
• The country produces around 77% of  the world’s EV batteries, according to reports as of 2022.
• China’s domination in the solar industry is even more daunting. It installed 329 GW of solar capacity in 2024, which was 55% of the total global capacity that year.

After being left behind in critical physical industries, the US can only play catch-up now, which is already a steep uphill climb. Deploying Physical AI might be the only way the US can get close to China. But that isn’t an easy task in itself. 

For instance, Marques Brownlee, a popular tech reviewer, discussed NEO, the most futuristic robot available in the US today. Although the company markets it as a smart robot, as things stand now, it’s anything but smart and autonomous. It works largely on human command and only does a handful of tasks without instructions. 

Brownlee noted that ‘AI companies’ are currently only selling the dream, before even fully creating the product. They’re using early-stage adopters as beta testers, but the gap between the blueprints and the final product is a huge one right now.

China is already deploying humanoid robots across frontline industrial applications. For instance, the Walker S2 from UBTECH, China’s most advanced humanoid robot, is capable of running 24/7 and can even replace its own battery. 

Compare this with NEO’s 4-hour battery life and teleoperation needs, and you’ll realize that the gap is wider than most Americans would want to believe. UBTECH plans to hit an annual capacity of 10,000 Walker S2 by 2027 for industrial use in smart factories, intelligent logistics, and automotive manufacturing.

Bezos’ Mission Critical Role

If there’s one man fit to lead such a herculean effort, it’s Jeff Bezos. His portfolio and experience are something even the brightest modern-day entrepreneur envies.

Bezos built the world’s largest logistics network, the most advanced warehousing system, a rocket company in Blue Origin, a cloud computing behemoth in AWS, and a robotics powerhouse in Amazon Robotics. That’s a solid CV.

Prometheus wants to build advanced physical stimulators, industrial-grade AI systems, and engineering automation tools, something that Bezos has dabbled in for 30 years.

Most importantly, he brings in the huge capital required by such a research-intensive operation. In addition to spending money out of his own pocket, the Bezos branding also helps attract top investors in the country, makes it easier to recruit top talent, and reassures investors about the viability of the mission.

Vik Bajaj, the other co-CEO of Prometheus, is also a very well-thought-out selection for the job. At a time when most AI leaders come with a background in machine learning and natural language processing, Bajaj comes from an experience in bioinformatics, biophysics, and molecular simulation.

He was the co-founder of Google’s Life Sciences division (which we now know as Verily), which is all about navigating huge R&D gestation periods and integrating robotics, sensors, AI, and cloud systems.

While Bezos is the face of the project, Vik Bajaj brings the hands-on experience and technological know-how required to tackle a business as large as this.

Unlike LLM-based AI startups, results for the physical AI space don’t come within just six months or a year. The R&D and trial-and-error phase could last for nearly half a decade, if not more.

A business this capital-intensive with a long gestation like this requires the support of someone who’s been there before – and both Bezos and Bajaj fit the bill perfectly.