I don't know if OpenAI's necessarily too big to fail. It could simply evolve and make a few changes - switch back to its original goal, fire everyone at the company who has spent more than an afternoon on Stanford's campus - and probably succeed with burning a lot less cash.

Remember, the original idea for OpenAI was to make open (thus the name) AI models that could only generate a maximum return of 100x for the investors. There was none of this hyperscaling, proprietary technologies, pure profit motive, etc. until later on.

I think it ultimately comes down to ideology. Simple financial math has nothing to do with it and never has. SV and DC, at least today, have a lot of beliefs in common. It can really be summed up as "divine right of CEOs". These are the "best and brightest" who came from the right parts of the country, who went to the right schools (even if they didn't finish their degrees), and ran in the right circles. They don't simply receive the opportunity to make unfathomable wealth and power; they're owed it. If they do something, by default, it must be right, and if a circumstance comes in their way, it's the circumstance that is wrong.

That's why you don't see a small motorcade of black Chevy Suburbans headed to these companies' headquarters from the local DHS field office. They aren't wrong; the expectation that the AI agent doesn't hack into government websites is wrong.

This is going to make for a great documentary in 5 years.