While some of the recent hype around AI may be due to VCs and the industry looking for the next big thing, it’s a bit absurd to think that AI orgs like OpenAI and Stability were holding back research and then only released it once the crypto bubble popped.
This article also misunderstands the differences between the various types of GPUs and seems to think they are fungible.
Large models like GPT-3 and Stable Diffusion are trained using cloud compute, most likely clusters of Nvidia A100s. No one was ever using those for mining because it wouldn’t have been profitable. Similarly, no large models are being trained on second hand GPUs used for mining.
This article also misunderstands the differences between the various types of GPUs and seems to think they are fungible.
Large models like GPT-3 and Stable Diffusion are trained using cloud compute, most likely clusters of Nvidia A100s. No one was ever using those for mining because it wouldn’t have been profitable. Similarly, no large models are being trained on second hand GPUs used for mining.