Oh man, I had literally this exact idea too (although targeting general GPU workloads, not AI specifically). I actually wrote a pretty detailed product roadmap/desiderata/business plan/basic market research/etc note about it.
I ended up not pursuing it to work on a different idea, feeling like Golem would probably eat that lunch. I love the simplicity of your approach - very in the spirit of an mvp. Let me know if you'd like to see the note (teaser: an ad that says "Because not everyone has a 500 GPU cluster", but "not" is crossed out and replaced with "now", over a background image of a Go board).
Anyways, best of luck, I'll be excited to see how it turns out!
I ended up not pursuing it to work on a different idea, feeling like Golem would probably eat that lunch. I love the simplicity of your approach - very in the spirit of an mvp. Let me know if you'd like to see the note (teaser: an ad that says "Because not everyone has a 500 GPU cluster", but "not" is crossed out and replaced with "now", over a background image of a Go board).
Anyways, best of luck, I'll be excited to see how it turns out!