I don't think AI will be able to sustain cloud compute-size margins. Inference is a commodity and value-adds can't be sold for anything. Why pay for an advanced harness when you can just ask the model to build one for you?

Observing SaaS vs using an open source stack in the beforetimes, the answer might be some combination of, "So I never have to think about it, and if anything goes wrong, no one can blame us because we went with the biggest & most trusted provider."