# Models mature onto consumer gaming GPUs

- **Speaker:** Greg Osuri
- **Attribution:** attributed
- **Said:** 2023-06-15
- **Theme:** Local Compute — https://www.gregosuri.com/predictions/local-compute/
- **Source:** RNDR, Supercloud, 1.5T AI Market Cap – Interview with Greg Osuri about Akash (Design DAO)
- **Watch at:** 00:41:49 — https://www.youtube.com/watch?v=CCHnfFIavA8&t=2509s
- **Canonical:** https://www.gregosuri.com/predictions/local-compute/2023-06-15-models-mature-onto-consumer-gaming-gpus-hkpw/
- **Cite as:** Greg Osuri, "Models mature onto consumer gaming GPUs," RNDR, Supercloud, 1.5T AI Market Cap – Interview with Greg Osuri about Akash (Design DAO), 15 Jun 2023, 00:41:49. https://www.youtube.com/watch?v=CCHnfFIavA8&t=2509s

## Quote — verbatim

> Most advanced models like GPT-4, 3.5... work well on high-end GPUs, but as they mature, as they get older, they get optimized to work on a distributed set of GPUs like 3090s, 4090s and whatnot — the lower-end GPUs, like gaming machines. So people that prefer these mature models over advanced models can use the lower-end GPUs.

## Context — site annotation, not spoken

Early articulation of his thesis that AI workloads migrate down to consumer hardware — frontier fine-tuning needs H100s (minutes vs hours), but mature models run distributed on gaming cards.