Prometheus raises $12B to build physical AI engineer

Prometheus, the physical-AI company co-founded by Jeff Bezos and Vik Bajaj, raised $12 billion in a new funding round that values the company at $41 billion, according to TechCrunch, CNBC, Semafor, The Next Web and other outlets. The round included institutional backers identified by reporting as JPMorgan Chase, Goldman Sachs, BlackRock, DST Global and Arch Venture Partners, and Bezos is reported to have invested personally (TechCrunch; Semafor). The startup previously announced an initial $6.2 billion raise when it launched late last year, and combined funding now exceeds $18 billion, per TNW and other reporting. Multiple outlets report Prometheus currently employs about 150 people across San Francisco, London and Zurich (TechCrunch; CNBC; TNW).

Editorial analysis - technical context: Public reporting describes Prometheus as building what Bezos and Bajaj call an "artificial general engineer," a class of models and systems intended to automate design and manufacturing workflows across complex physical domains such as aerospace, advanced manufacturing and drug discovery (TechCrunch; Semafor; TNW). Reporting says the company trains models on experimental and industrial data, combines physics knowledge with empirical testing results, and operates a large internal GPU cluster while also buying external compute (Semafor; CNBC). Sources say the team has been hiring from major AI labs and industry engineering teams but has kept concrete product demonstrations and technical benchmarks private so far (CNBC; TechCrunch).

The size of this raise and the emphasis on physical-world engineering echo a recent investor shift toward "physical AI," where founders and backers argue that real-world data and manufacturing workflows create defensible moats compared with software-only models (TechCrunch; TNW). Reporting frames Prometheus as one of the largest private bets on that thesis, with public-figure involvement and scale of capital that are unusual even in this funding cycle (TechCrunch; Semafor). Observers in coverage note parallels to prior cases where large compute commitments and proprietary data were necessary to advance domain-specific models, for example in biotech and robotics (TechCrunch; Semafor).

In an exclusive interview with CNBC, Bezos described the work as "very compute intensive" and said that creating the data necessary for the models is a major reason for the large funding needs. Semafor and TNW reported that Bajaj said some data comes from collaborations with companies improving their processes, and both founders declined to disclose many technical specifics. Semafor and The Next Web report the founders discussed, but would not fully detail, a broader vision that could involve acquiring companies to apply Prometheus technology in industry.

For practitioners: Large, compute-heavy efforts that aim to close the gap between simulation, experiment and production typically require multi-disciplinary stacks-data acquisition pipelines from physical tests, high-fidelity simulation, control and robotics integration, and industrial validation workflows. Reporting indicates Prometheus is investing across those areas, but there are no public benchmarks or reproducible artifacts available yet (CNBC; TechCrunch; Semafor). Teams evaluating partnerships or recruitment should expect long horizons for prod-ready, domain-validated models in this class, based on prior examples in industrial AI.

For observers: watch for any public technical papers, reproducible benchmarks, open datasets or third-party validations that demonstrate generality across multiple physical-engineering tasks. Also monitor compute procurement disclosures and partnerships with manufacturing or pharma firms, which reporting identifies as likely sources of training data and validation but which the founders have not fully named (Semafor; CNBC). Finally, funding scale and investor mix mean the company will be an influential signal for capital flow into physical-AI startups; coverage suggests additional fundraising or M&A activity tied to this vision may follow, though the founders have not provided a public roadmap (TNW; Semafor).