The latest DeepSeek and Kimi open weight models are competitive with GPT-5.

If every AI lab were to go bust tomorrow, we could still hire expensive GPU servers (there would suddenly be a glut of those!) and use them to run those open weight models and continue as we do today.

Sure, the models wouldn't ever get any better in the future - but existing teams that rely on them would be able to keep on working with surprisingly little disruption.