Jeff Bezos seeks $100 billion industrial AI fund, Project Prometheus ties buyouts to factory automation, software ambition shifts from selling tools to owning bottlenecks

Images

Image Credits:David Ryder / Getty Images techcrunch.com

DoorDash app on phone techcrunch.com

techcrunch.com

Jeff Bezos is seeking to raise a $100 billion fund to buy established industrial companies and retrofit them with AI-driven automation, according to TechCrunch citing reporting by The Wall Street Journal. The effort is tied to Project Prometheus, an AI startup Bezos is said to co-found and co-lead with former Google executive Vik Bajaj, after an initial $6.2 billion funding round. The acquisition targets described include aerospace, chipmaking and defense—sectors where production capacity, certification, and supply-chain access are hard to replicate quickly.

The pitch is that “old” manufacturing can be made dramatically more productive by applying new models to engineering and factory operations. But the structure described—an AI company paired with a buyout vehicle—also changes who captures the gains. Instead of selling software to many factories, the fund would buy the factories and deploy Prometheus’ tools internally, turning AI from a vendor product into a control layer over physical bottlenecks. In industries where lead times run in years and qualified suppliers are scarce, owning the asset can matter more than having the best model.

That approach resembles private equity in its mechanics—acquire, standardise, cut costs, and improve margins—but with a different asset at the centre: data. Once a fund controls multiple plants across a sector, it can centralise process data, train models on proprietary workflows, and spread optimisation playbooks across a portfolio. The resulting advantage is less about building new capacity than about making existing capacity cheaper and more predictable, then using that predictability to win contracts.

It also shifts the risk profile. Industrial modernisation is usually constrained by downtime, safety regulation, and labour agreements; the expensive part is not writing code but changing a running system without breaking it. A large pool of committed capital can absorb those transition costs—and, if the plan works, the fund can sell the story as “AI-enabled productivity” even when the hard work is procurement discipline, maintenance schedules, and tighter supervision.

Bezos has been travelling to Singapore and the Middle East to raise money for the project, according to the Journal’s sources. Those regions have deep pools of capital looking for exposure to strategic industries, and they have shown a willingness to finance long-duration bets when the assets are tangible.

For now, the proposal is still fundraising and target selection. The first test will be whether Prometheus can turn a patchwork of factories into a single machine that produces both parts and data on schedule.