I genuinely think it's part of a psyop. If we bloat all codebases and eventually start printing the models on chips to reduce inference costs by 50-100x they'll

I genuinely think it's part of a psyop. If we bloat all codebases and eventually start printing the models on chips to reduce inference costs by 50-100x they'll take in massive profits from 5M line codebases instead of 350k