> It's nobody gets fired for buying IBM all over again.

I think that's the historical analogue. How is IBM doing compared to pre-personal-computer disruption? Initially-limited home OSes like DOS were good enough to eventually dominate business too. The AI labs, with their massive funding and spending, are speedrunning the whole thing in a way that just might make that disruption faster and more fatal vs the lingering zombie relying on the IBM name. (The more massive amounts of capital you raise on future speculation of enormouse TAMs before, say, becoming profitable, the more dependent you are on the future speculations of outside interests. Double-edged sword.)

And I don't think being suspicious of the future of OpenAI margins is the same as saying open source DIY will dominate at all.

People don't wanna install their own OS or deal with changes in their office suite or rack their own servers, but a low-cost AI provider is gonna be more like a Wikipedia-vs-Encarta situation in terms of accessibility and similarity-of-interface.