Jeff Bezos has backed a new AI startup called Prometheus with $12 billion in funding to develop what it calls an artificial general engineer, a system capable of performing the work of a human software engineer across complex, real-world tasks.
The round, one of the largest in AI startup history, positions Prometheus as a direct challenger to established research labs building general-purpose reasoning systems. The funding comes from a consortium of investors led by Bezos, with participation from several large sovereign wealth funds. The company has not disclosed a valuation, though the scale of the raise implies a figure well into the tens of billions.
Unlike systems focused on language generation or narrow task completion, Prometheus is building toward what its founders describe as an artificial general engineer, or AGE: a model that can plan, debug, test, and deploy code with minimal human oversight. The goal is not writing isolated code snippets, but handling full engineering projects from specification to production deployment.
Speaking to CNBC, Bezos offered a measured take on the rationale: “Significant productivity in the economy is going to raise the standard of living.” He stopped short of specific timelines but suggested the economic impact of capable AI engineers would be felt broadly across industries, not just in software.
The timing is notable. AI coding tools have multiplied rapidly over the past three years, from GitHub Copilot to Cursor to Devin. Most remain assistants that augment developers rather than replace them. Prometheus is betting the next leap requires fully autonomous engineering at scale, not incremental productivity gains layered on top of existing workflows.
Prometheus has not disclosed technical details about its model architecture, though the company published a research paper describing its approach to long-horizon reasoning and multi-step planning. The system reportedly handles tasks that require holding context across hundreds of files and thousands of lines of code, a capability that current general-purpose models handle inconsistently.
The $12 billion figure stands out even in a market used to large AI rounds. For comparison, OpenAI’s most recent raise was $6.6 billion and Anthropic raised $7.3 billion. Prometheus has not released any product accessible to outside customers, which means the valuation rests entirely on the team and the research direction.
That team is the company’s clearest credential. Several Prometheus co-founders previously led major research groups at Google DeepMind and Meta AI, giving the startup immediate credibility in a field where research pedigree carries significant weight with both investors and potential enterprise customers.
For developers and technology companies watching the space, the central question is whether Prometheus can deliver a meaningful capability advance before well-funded competitors close the gap. With $12 billion in hand and Bezos as a public backer, the company at least has the runway to find out.
