“Irrational exuberance.”

There are parallels here of folks getting so caught up in the hype that they forgot business fundamentals. Everytime folks say “but this is different” and every time it’s not.

It's certainly different this time; machines have never been able to do anything like what they can do now. You might as well argue that the Industrial Revolution or the Internet itself wasn't "different this time."

But the phrase "this time" requires a lot of hand-waving. The current generation of models is obviously a bubble, which means that businesses based on them are also participants in a bubble. The market seems to be pricing in various unspecified future miracles. Given the history of AI to date, the miracles needed to justify current valuations might arrive next week, next year, or 30 years from now.

They will arrive, though. That part is no longer up for debate by anyone who's been paying attention since AlphaGo, never mind Vaswani.

I don't think anyone is questioning current or future capabilities.

It's more of a question if it will ever actually be profitable or marketable without subsidizing most of the cost of running it.

We're seeing the same with streaming services right now. 5-10 years ago, everyone thought they needed their own streaming service and heavily invested into building one and acquiring licenses or producing content for it. Now we're seeing the part where they are trying to make it profitable by raising prices/adding ads or both.

I'm not sure if OpenAI will ever be able to just run by themselves. Without major outside investments to subsidize the cost of actually having users use their services.

But that isn’t different than previous bubbles, either. The first dot com crash was an over investment in online shopping. And yet we all online shop for everything now. Same with the video games crash in the 80s, now a wildly successful industry.