JERA and UC Berkeley team up on physical AI and energy innovation

This MOU quietly links one of Japan’s largest power producers with a top US research university around a portfolio that explicitly includes “physical AI” alongside decarbonization and digital technologies. JERA brings capital, large‑scale infrastructure and global energy market exposure; Berkeley brings robotics, AI and commercialization pipelines. The collaboration is structured to go beyond papers: it emphasizes startup commercialization, internships and using JERA Ventures as an implementation hub.([prnewswire.com](https://www.prnewswire.com/news-releases/jera-signs-mou-with-uc-berkeley-to-advance-collaboration-in-education-science-and-energy-innovation-302733898.html))

For the AGI race, this is another example of how frontier‑adjacent AI—here, robotics and control in energy systems—is being woven into national‑scale infrastructure. If physical AI can meaningfully improve grid stability, asset inspection, or flexible demand response, it will unlock more capacity for power‑hungry data centers and supercomputers, indirectly supporting frontier model training. It also suggests that energy incumbents want a seat at the table on how embodied AI is deployed in critical infrastructure, not just tech firms.

The cross‑border nature of the deal matters too. In an era of tightening export controls, academia‑industry partnerships like this become key channels for sharing techniques and building shared norms around safe deployment. If done well, JERA‑Berkeley could become a reference for how to align physical‑AI innovation with climate goals rather than just efficiency metrics.