Project Prometheus targets massive capital infusion to scale physical world AI models.
The AMW Read
Introduces a new top-tier physical AI player with a massive $10B raise and $38B valuation, signaling a major capital-driven shift toward vertical integration of industrial data.
Project Prometheus targets massive capital infusion to scale physical world AI models.
Jeff Bezos's AI startup, Project Prometheus, is finalizing a $10 billion funding round involving major financial institutions such as JPMorgan and BlackRock. This follows an initial $6.2 billion capital 확보 in November of last year, propelling the company's valuation to an estimated $38 billion. Led by co-CEO Vikram Bajaj, the startup is positioning itself as a dominant force in the physical AI sector. The company is also establishing a separate investment holding company to deploy billions of dollars into engineering, design, and manufacturing firms to secure a proprietary data-investment flywheel.
The rapid valuation surge places Project Prometheus ahead of other significant players like Ilya Sutskever's SSI, which recorded a $32 billion valuation in April 2025. This development signals a market shift toward 'World Models'—AI designed to understand physical laws and automate complex industrial processes. By targeting sectors like aerospace, automotive, construction, and manufacturing, the company is moving beyond LLMs toward high-stakes industrial automation. The recruitment of high-level talent, including former OpenAI infrastructure lead Kyle Koch, underscores the intense competition for engineers capable of building massive-scale computing infrastructure for physical reasoning.
The strategic move to invest in manufacturing companies for data acquisition represents a sophisticated approach to the data scarcity problem facing frontier AI labs. Rather than relying solely on internet-scale text, Project Prometheus is building a vertical integration strategy where capital investment drives the creation of high-quality, specialized industrial datasets. This 'data-investment flywheel' could create significant barriers to entry for competitors attempting to enter the physical AI space. As the company expands its footprint across San Francisco, London, and Zurich, its ability to merge large-scale infrastructure with real-world physical processes will likely define the next frontier of industrial intelligence.
