# Centralized AI training hits an energy cap in two years

- **Speaker:** Greg Osuri
- **Attribution:** attributed
- **Said:** 2025-03-17
- **Theme:** Energy & AI — https://www.gregosuri.com/predictions/energy-ai/
- **Source:** Why the Future of AI Depends on Decentralized Cloud Platforms (Eye on AI)
- **Watch at:** 00:55:53 — https://www.youtube.com/watch?v=Rq0UByMDUv4&t=3353s
- **Canonical:** https://www.gregosuri.com/predictions/energy-ai/2025-03-17-centralized-ai-training-hits-an-energy-cap-in-two-n6te/
- **Cite as:** Greg Osuri, "Centralized AI training hits an energy cap in two years," Why the Future of AI Depends on Decentralized Cloud Platforms (Eye on AI), 17 Mar 2025, 00:55:53. https://www.youtube.com/watch?v=Rq0UByMDUv4&t=3353s

## Quote — verbatim

> Using synthetic data and using mixture of experts mechanism you can actually solve the data problem, but what we cannot solve is the energy problem. That's why it's very, very important, if you're doing training, to focus on distributing your training runs versus trying to go with the traditional mechanism of centralizing your training runs... because we're going to hit a cap in two years and we have no solutions.

## Context — site annotation, not spoken

His closing thesis (also the cold open): DeepSeek proved synthetic data removes the data ceiling, leaving energy as AI's binding constraint; $500B in announced investment goes to power infrastructure that will arrive too late.