Do you have any evidence that inference revenue is growing faster than training costs? RLVR is significantly less compute-efficient than token-prediction pretraining - especially as labs are trying to train models to achieve agentic tasks which take tens of minutes per rollout.

I don't have any evidence. You'll have to believe what Anthropic and OpenAI CEOs say publicly.

However, it seems to make a lot of sense. Anthropic literally added $6b ARR in February 2026 alone. I doubt training costs go up that fast.