Compute that follows green energy: train in Arizona sun, infer on Kansas wind
Akash Accelerate 2025 - Greg Osuri Keynote (Akash Network)
“Workloads can migrate to where the power is greenest, cheapest, and most available. When it’s sunny in Arizona, we train there. When wind peaks in Kansas, we infer there. This isn’t a small optimization. It is a fundamental shift. It turns compute from static burden into a dynamic asset. Instead of straining the grid, it balances it.” — 00:16:05
Context: Argues decentralized AI infrastructure can be location-, time-, and energy-aware — “a new symbiosis between AI and clean energy.”