Globally distributed consumer-GPU clusters viable within two years
Building the Super Cloud of GPUs with Akash Founder Greg Osuri | EP #102 (Frictionless Podcast by Logan Jastremski)
“Right now it’s not the most optimal setup to leverage a globally distributed clusters of 4090s, but in two years I bet you is going to be the case — and that’s really where we’re heading.” — 00:41:51
Context: Explains cost-of-compute vs. cost-of-communication tradeoff; argues research has optimized for local clusters only because AI is pre-productionization, and maturing open models (Llama 2, Alpaca on 3090s) will unlock distributed topologies.