# In two years, training a frontier model will need a nuclear reactor

- **Speaker:** Greg Osuri
- **Attribution:** attributed
- **Said:** 2024-11-25
- **Theme:** Energy & AI — https://www.gregosuri.com/predictions/energy-ai/
- **Source:** Greg Osuri \| Trump's impact on crypto x AI, why DePIN is inevitable, and Akash Network revenue ATH's (Proof of Coverage Media)
- **Watch at:** 00:22:29 — https://www.youtube.com/watch?v=Rp_LL6tPDFU&t=1349s
- **Canonical:** https://www.gregosuri.com/predictions/energy-ai/2024-11-25-in-two-years-training-a-frontier-model-will-need-a-j66f/
- **Cite as:** Greg Osuri, "In two years, training a frontier model will need a nuclear reactor," Greg Osuri \| Trump's impact on crypto x AI, why DePIN is inevitable, and Akash Network revenue ATH's (Proof of Coverage Media), 25 Nov 2024, 00:22:29. https://www.youtube.com/watch?v=Rp_LL6tPDFU&t=1349s

## Quote — verbatim

> The next big LLM will need \[a\] 250 megawatt data center, and a year after that is going to be 1.2 gigawatt data center, and the only place to get 1.2 gigawatts of capacity is a nuclear reactor. So in two years we're going to need a new nuclear reactor to train the most powerful state-of-the-art model, and I think that is a huge bottleneck we don't talk much about.

## Context — site annotation, not spoken

Extrapolating from GPT-4's ~10MW training data center; he notes the last US reactor (Georgia) cost ~$32B and took ~14 years, so optimistically seven years to build.