# By 2030, training frontier models requires gigawatt supply — nuclear is the only way

- **Speaker:** Greg Osuri
- **Attribution:** attributed
- **Said:** 2025-07-05
- **Theme:** Energy & AI — https://www.gregosuri.com/predictions/energy-ai/
- **Source:** Akash Accelerate 2025 - Javier Villamizar & Greg Osuri Fireside chat (Akash Network)
- **Watch at:** 00:07:25 — https://www.youtube.com/watch?v=F_wsWsZTaxs&t=445s
- **Canonical:** https://www.gregosuri.com/predictions/energy-ai/2025-07-05-by-2030-training-frontier-models-requires-gigawatt-exax/
- **Cite as:** Greg Osuri, "By 2030, training frontier models requires gigawatt supply — nuclear is the only way," Akash Accelerate 2025 - Javier Villamizar & Greg Osuri Fireside chat (Akash Network), 5 Jul 2025, 00:07:25. https://www.youtube.com/watch?v=F_wsWsZTaxs&t=445s

## Quote — verbatim

> Considering the energy need — particularly concentrated energy need for training — is doubling every two years, so the expectation by 2030 the only way to train the state-of-the-art model is with a gigawatt supply. So nuclear will be the only way to do so. Unfortunately, the last nuclear reactor we built in America took about 14 years, and the interconnect requests now average anywhere from 10 years here for new data center builds.

## Context — site annotation, not spoken

Greg's setup to a policy question about reducing interconnect queues and regulatory burden.