7 comments

[ 0.29 ms ] story [ 4.6 ms ] thread
(comment deleted)
this data is from June
This is data from 2025 ... I doubt 2025 has changed materially since June 2026.
Does it matter? I don't think anyone expects the current LLM generation to be profitable in the next few years. The bet I think is that OpenAI will be the first to develop ASI or at least among the leaders of the race.
It matters because it's hard to see getting profitable enough to pay for all that CapEx.

- OpenAI can't create moat like current SaaS companies. It can't benefit from monopoly pricing. The business will be highly competitive.

- Industry will be capital heavy. Constant investments into R&D and new hardware every 4 years make it look like old industry with heavy physical 'factories'. Huge revenues small profits.

Users will benefit, just like they will benefit from telecoms, railway or web booms, but most early birds will go bust and will be bought out.

The bet is the "total addressable market" that is: theres enough demand to payout trillions and trillions.

The type of AI is arbitrary. The money is to create a huge dependency layer and its a huge drlusion youre seeing as data center fights.

In other words, if AI really demands these values, what are the product and will there actually be buyers at that price point. The singularity then is not about self improving AI, BUT self improving capitalism where one business controls all the resources and one person consumes them all and the rest of use pray for trickle down economics.