[postalrat]

Access is limited by the hardware you have access to. Forget about training, just running these models takes a lot of memory.

[mcbuilder]

AI compute is increasing at a rate faster than Moore's law. Right now if DALL-E 2 takes an 8xA100 box to run inferences, how long before it's on the workstation? 4-5 years?

[selfhoster11]

It's to be expected. Almost nobody really cared about accelerating AI computations, so I expect that AI chips will be catching up.