According to reports citing informed sources, Jeff Bezos, the billionaire founder of e-commerce and cloud computing leader Amazon.com, is seeking to leverage cutting-edge AI technology to drive a super-revolution in manufacturing. He is currently in discussions with major Wall Street financial institutions to raise $100 billion to establish a large-scale fund. This fund would be used to acquire manufacturing companies worldwide and infuse them with artificial intelligence technology. The concept of "AI integrating into everything" is no longer just a slogan but an industrial-scale transition already underway, particularly irreversible in real-world sectors such as manufacturing, logistics, robotics, defense, data centers, and industrial control. By raising such a massive amount to deploy AI in manufacturing, Bezos is essentially betting that the next phase of AI will not merely sell computing power and models but will directly transform factories, equipment, large-scale enterprise project processes, and asset returns.
Reports indicate that in recent months, Bezos has met with leaders of several major asset management firms to secure funding for this project, including numerous private discussions during his trips to the Middle East and Singapore. A spokesperson for Bezos did not immediately respond to requests for comment. Last November, multiple media outlets reported that Bezos, who stepped down as CEO of Amazon.com in 2021 but remains chairman of the retailer, co-founded and provided the majority of funding for Project Prometheus. This is an AI technology company focused on applying artificial intelligence in the real physical world. According to employee LinkedIn profiles, the AI firm, co-founded by former Google veteran Vik Bajaj and Bezos, has not publicized its ambitions as prominently as Musk's xAI but has been recruiting AI talent in San Francisco, Zurich, and London.
The battle for capital focused on AI in the physical world has begun. Project Prometheus aims to reshape the physical world using AI. Reports suggest Bezos is involved in the management of this AI technology company as co-CEO, signaling the high-profile return of the former world's richest man and Amazon.com founder to the global tech hub of Silicon Valley. Insiders note that Bezos is committed to positioning Project Prometheus as one of the leading AI tech companies focused on applying artificial intelligence to cutting-edge physical AI tasks. These include large-scale advanced manufacturing projects, humanoid robotics, drug design, and exploration and discovery in physical sciences—highly oriented toward physical AI applications such as manufacturing, automation, equipment control, and real-world scenario optimization. This also helps explain why he is now seeking to raise enormous funds to acquire manufacturing firms and integrate AI. In other words, the underlying goal of Project Prometheus is not to create another pure-software large-model AI startup like OpenAI, but rather to turn AI into an "industrial operating layer" that transforms real production systems.
This marks the first time Bezos has taken a formal, senior operational role at a company since stepping down as Amazon.com CEO in July 2021. Although he has been deeply involved in managing space company Blue Origin—the strongest competitor to Tesla CEO Elon Musk's SpaceX—his official title at Blue Origin remains "founder." In the realm of AI applications, Musk has his own AI startup, xAI, attempting to rival OpenAI's ChatGPT with its Grok AI. According to insights from NVIDIA CEO Jensen Huang, "physical AI" emphasizes enabling robots/autonomous systems to perceive, reason, and act in the real world. An era where "physical AI" assists the evolution of human civilization is approaching. These three capabilities are key tools to advance models from "just conversing" to "working in the physical world."
If Bezos's co-founded and funded Project Prometheus is to establish a lasting AI barrier, its most valuable asset may not be singular model capability, but its ability to integrate perception, reasoning, control, simulation, manufacturing process data, and real-world feedback into a closed loop. Whoever first embeds AI into production lines, equipment, robots, experimental systems, and engineering decision-making processes will be closer to "completely reshaping the physical world with AI." Bezos's plan to set up a $100 billion fund to acquire manufacturing companies and inject AI is not an isolated event; it aligns with the same vision as his co-founded and wholly focused "AI in the physical world" company, Project Prometheus.
Meanwhile, the world's current richest man, Musk, is also pushing AI from the model layer to the physical system layer. Tesla is advancing its next-generation self-developed AI chip—the AI6 chip—and has launched a plan to build a massive AI chip superfactory. Both Tesla and SpaceX AI (SpaceX and xAI have merged) have stated they will continue large-scale purchases of NVIDIA's high-performance AI computing infrastructure. At the latest GTC conference, NVIDIA CEO Jensen Huang significantly raised the revenue opportunity for NVIDIA's AI computing infrastructure between 2025-2027 to at least $1 trillion—far above the previous GTC blueprint of $500 billion by 2026—and clearly pointed to the next wave of opportunity in agentic AI and physical AI.
These signs indicate that the world's wealthiest and most resourceful individuals are upgrading AI from "merely generating辅助型 content" to a "productivity transformation system that integrates AI into everything and reconstructs the real physical world." The recent strengthening of this "AI integrating into everything" trend in the physical world is not solely due to capital enthusiasm but because technological conditions have begun to form a closed loop: large models handle perception, understanding, and reasoning; edge computing enables low-latency execution; embodied AI like "AI+robotics" and automation equipment provide "hands and feet"; digital twins and simulation offer training grounds; and data centers continuously drive down the cost of training and inference systems.
A typical example is Skild AI and NVIDIA deploying a general-purpose robot "brain" on Foxconn's Blackwell AI server rack production line, using Skild AI-built robots to produce NVIDIA server racks. NVIDIA itself released the Open Physical AI Data Factory Blueprint for robotics, visual agents, and autonomous driving at GTC and is advancing its physical AI ambitions with several robotics companies. In other words, AI is no longer just sitting in the cloud answering questions but is beginning to enter real-world physical processes like assembly,搬运, inspection, scheduling, and control.
Comments