HomeArticle

Jeff Bezos has once again "copied" Elon Musk's playbook.

新智元2025-11-18 14:32
Bezos is back in the game, targeting physical AI! Musk bluntly said, "Stop copying my work!"

Before Grok 4.1 even made its debut, the tension in Silicon Valley has been completely ignited by Bezos!

However, Musk can't stop laughing:

Jeff Bezos, the founder of Amazon and the world's wealthiest billionaire, has started "copying Musk's homework" again.

After four years, Bezos has donned the CEO mantle again. Together with Vik Bajaj, the co-founder of Google's life sciences department and Verily, a biotech company under Alphabet, he founded the AI company "Project Prometheus", shifting his focus from the internet to "Physical AI".

Different from the current mainstream large language model companies, "Project Prometheus" aims to build an AI model that can learn by analyzing physical processes such as robot experiments, with application areas including computers, automobiles, and aerospace.

Musk said, "This is all copying me!":

Previously, he said that Tesla will be an AI company in the future.

He even bluntly stated that the only way to save the US economy is artificial intelligence (AI) and robots.

Currently, Bezos' startup has poached nearly a hundred researchers from giants like OpenAI and Meta. The first-round funding is as high as $6.2 billion, far exceeding most players in the same field, attempting to repeat Amazon's infrastructure story.

This move is not only a new start in Bezos' career but may also rewrite the power map of the real - world industry and the AI competition.

Did Bezos Copy Musk's Homework Again?

In the past few years, Bezos' focus has not been limited to business - his personal life has also attracted outside attention.

This year, he held a star - studded and luxurious wedding in Venice, which sparked a lot of discussion.

At the same time, he has been more involved in Blue Origin and has shown an increasing interest in the AI field.

However, it seems that Bezos always follows Musk's steps closely:

Musk created rocket launches into space --> Bezos created rocket launches into space

Musk's SpaceX made rockets land --> Bezos' "Blue Origin" made rockets land

Musk said he wanted to go to Mars --> "Blue Origin" launched NASA's Mars artificial satellite

Musk created the best AI model --> Bezos created an AI model

No wonder netizens say that someone is "copying" Musk's "wealth - making recipe".

Netizen Valentina Gomez "suggested" that Bezos should change his last name to Musk and be called Jeff Musk:

Bezos' new company, Project Prometheus, has remained low - key so far, and even the specific founding time has not been disclosed to the public.

According to multiple sources, the core direction of this startup exactly fits Bezos' long - standing "space dream" -- focusing on the application of artificial intelligence in engineering and manufacturing fields such as computers, aerospace, and automobiles.

Sharing the position of co - CEO with Bezos is physicist and chemist Vik Bajaj. He once worked at Google X ("Moonshot Factory") and closely collaborated with Google co - founder Sergey Brin, participating in the incubation of many radical innovations such as the drone delivery service Wing and the self - driving car project Waymo.

In 2015, Bajaj co - founded the life science research institution Verily (which, like Waymo and Wing, is under Google's parent company Alphabet).

Three years later, he founded Foresite Labs, dedicated to incubating startup projects in the fields of AI and data science.

Recently, he has stepped down from that position and is fully committed to Prometheus.

So, why do these billionaires take a fancy to real - world AI including robots?

The New Competition of "AI × Real World"

In recent years, a wave of "physical intelligence" has emerged in the United States, and Project Prometheus is just one of them.

These companies are not satisfied with keeping AI only at the text - processing level but are focusing on complex tasks in the real world -- from robot operations, drug research and development to scientific discovery.

For example, several researchers who left big companies like Meta, OpenAI, and Google DeepMind founded Periodic Labs, aiming to accelerate scientific research in the fields of physics and chemistry through physical simulation experiments.

Periodic Labs has received $300 million in financing and plans to build a laboratory in northern California. It will use robots to run scientific experiments on a large scale, allowing AI to learn from continuous trial - and - error, and ultimately achieve "self - experimentation".

Bezos clearly also sees the potential in this direction.

As early as last year, he invested in the robot AI company Physical Intelligence.

Now, Prometheus has $6.2 billion in startup funds, far exceeding most competitors -- even Thinking Machines Lab, founded by former OpenAI employees, only raised $2 billion this year.

According to sources, Prometheus currently has nearly 100 employees, including researchers poached from top AI institutions such as OpenAI, DeepMind, and Meta.

Established giants like OpenAI, Google, and Meta are also vigorously deploying in the direction of "AI for Science". For example, two researchers from Google DeepMind won the Nobel Prize in Chemistry for the AlphaFold protein structure prediction project.

But Prometheus and emerging companies like Periodic Labs are taking a more ambitious path:

They hope to train AI models that can not only understand text but also "read" the laws of the real world.

This goes beyond the learning mode of existing large language models (LLMs) --

LLMs rely on a large amount of text content on the Internet. By identifying language patterns in information such as Wikipedia, news, and blogs, they learn to "imitate human speech".

They can write code and do math problems, but they are just playing in the "language world".

What Prometheus wants to build is "physical AI" that can observe the world, conduct experiments, and continuously improve.

For AI to truly have general intelligence, it can't just "read thousands of books" but also "travel thousands of miles": conduct experiments in the physical world.

The Last AI Safe Haven?

At the "Italian Technology Week 2025" recently held in Turin, Italy, Bezos admitted in a fireside chat when talking about the current AI wave: "There are indeed signs of a bubble."

But he believes that when the trend fades, the remaining winners will bring far - reaching value to society.

When the dust settles, we will see who the ultimate winners are.

And their inventions will benefit the whole society... The benefits brought by AI will be huge.

Now, he has chosen to start a new business again, placing a big bet on "physical intelligence".

Physical AI refers to enabling automated systems such as cameras, robots, and self - driving cars to have the ability to perceive, understand, reason, and execute or coordinate complex tasks in the real world.

Put simply, it means allowing AI not only to "understand the world" but also to "take action" and truly participate in the operations and decision - making in the physical world.

Physical AI has become the next wave of AI.

For example, Jensen Huang said that the next wave of AI will be physical intelligence. In the future, factories will be robotized, and robots will command the production of robotic products.

Nvidia has already deployed in fields such as robots, self - driving vehicles, and intelligent spaces.

And Musk claims that he plans to deploy thousands of Optimus robots in Tesla factories by the end of 2025.

He also proposed a more radical production - capacity plan: by 2030, produce one million units annually.

I believe we can achieve an annual production of one million units in four years... At the latest by 2030, or even 2029.

Musk's vision of Optimus goes far beyond heavy physical work:

  • It can work around the clock, without pay, and without fatigue;
  • After being trained to complete various tasks, it will completely reshape the labor market and economic structure;

He even said: "Work will become a choice