Wouldn't enterprises rather run their own models locally?

While not today, very soon every company, of every complexity will run local models. It's not in a companies interest to hand over its domain expertise, data, and proprietary IP for a increase in productivity. Most will quickly realize it makes sense to run their own weights. This will be commonplace once tooling and training infrastructure is commoditized.

Enterprises don't even want to self-host webservers, and those are about a thousand times easier to do than self-hosting an AI model