# Distributed training becomes a thing in 6 months to a year

- **Speaker:** Greg Osuri
- **Attribution:** attributed
- **Said:** 2026-03-28
- **Theme:** Decentralized AI — https://www.gregosuri.com/predictions/decentralized-ai/
- **Source:** Akash Network - Decentralized Cloud Built for AI's Next Frontier (DePIN Connection)
- **Watch at:** 00:34:20 — https://www.youtube.com/watch?v=QuVl2mVHRyo&t=2060s
- **Canonical:** https://www.gregosuri.com/predictions/decentralized-ai/2026-03-28-distributed-training-becomes-a-thing-in-6-months-t-xvxn/
- **Cite as:** Greg Osuri, "Distributed training becomes a thing in 6 months to a year," Akash Network - Decentralized Cloud Built for AI's Next Frontier (DePIN Connection), 28 Mar 2026, 00:34:20. https://www.youtube.com/watch?v=QuVl2mVHRyo&t=2060s

## Quote — verbatim

> I feel like 6 months to a year's time we're going to see distributed clusters and distributed training going to be a thing because the technology is not there yet. I mean, we're kind of there, but not really there yet.

## Context — site annotation, not spoken

After walking through the three blockers (communication overhead, fault tolerance, heterogeneous GPUs) and praising Pluralis for solving two of them with a 7B distributed-trained model.