4 ms·
those gross profit margins aren't that useful since training at fixed capacity is continually getting cheaper, so there's a treadmill effect where staying in bu
by adgjlsfhk1 9mo ago
those gross profit margins aren't that useful since training at fixed capacity is continually getting cheaper, so there's a treadmill effect where staying in business requires training new models constantly to not fall behind. If the big companies stop training models, they only have a year before someone else catches up with way less debt and puts them out of business.
- HDThoreaun 9mo agoOnly if training new models leads to better models. If the newly trained models are just a bit cheaper but not better most users wont switch. Then the entrenched labs can stop training so much and focus on profitable inference
- kuschku 9mo agoIf they really have 40-60% gross margins, as training costs go down, the newly trained models could offer the same product at half the price.
- HDThoreaun 9mo agoWell thats why the labs are building these app level products like claude code/codex to lock their users in. Most of the money here is in business subscriptions I think, how much savings would be required for businesses to switch to products that arent better, just cheaper?
- kuschku 9mo agoI think the real lock-in is in "CLAUDE.md" and similar rulesets, which are heavily AI specific.
- yencabulator 9mo agoWhy would they be heavily "AI specific", when we're being told these things are approaching AGI and can just read arbitrary work documents?
- bombolo 9mo ago[dead]