Bezos Builds Prometheus To Modernize Manufacturing

Jeff Bezos and former Google Life Sciences executive Vik Bajaj are leading Project Prometheus, an AI startup that launched in November 2025 with roughly $6.2 billion in initial funding. Prometheus is assembling teams and capital to build high-level AI systems aimed at the physical economy — manufacturing, aerospace, automotive, chipmaking and defence — and is actively hiring engineers and infrastructure specialists. Financial and strategic moves include raising “tens of billions” and a reported push by Bezos to assemble a $100 billion acquisition fund to buy and modernize incumbent industrial firms.

Prometheus positions itself outside the chatbot-first narrative. Its engineering brief centers on industrial-scale, production-ready AI: models and stacks that integrate with factory automation, engineering workflows and domain-specific data pipelines. That requires deep systems work — compute and data infrastructure for large models, offline/edge integration, simulation and digital-twin engineering, and access to proprietary industrial datasets often held by existing firms.

Prometheus had hired roughly 100 employees across San Francisco, London and Zurich by last November, poaching researchers from OpenAI, DeepMind and Meta and prioritizing talent with experience building large infrastructure projects. FT-sourced reporting (via Silicon Republic) confirms Kyle Kosic — an xAI co-founder who later worked at OpenAI — has joined Prometheus to lead AI infrastructure efforts. TechCrunch and the WSJ connection describe Bezos’ effort to raise massive acquisition capital, targeting aerospace, chipmaking and other sectors, and engaging sovereign investors in Singapore and Gulf states. Prometheus plans to take equity stakes in target companies both to accelerate AI deployments and to secure domain training data.

This is a material shift in where deep learning and foundation-model engineering can be applied. Prometheus’ emphasis on industrial automation and acquisition-driven data access signals competition for enterprise-grade datasets, deployment engineering, and compute-integration talent. Expect increased demand for engineers who can productionize large models at the edge, integrate ML with PLCs and CAD/CAM toolchains, and design safe, verifiable agentic systems for physical processes.

hires and published papers or open-source infrastructure from Prometheus (indicating technical direction); announced acquisitions and revealed dataset access patterns; partnerships with sovereign and industrial partners; and early product integrations in aerospace, chipmaking or automotive lines.