Code red at OpenAI as it 'pours money down a black hole'(telegraph.co.uk)
telegraph.co.uk
Code red at OpenAI as it 'pours money down a black hole'
https://www.telegraph.co.uk/business/2026/03/29/code-red-at-openai-as-it-pours-money-down-a-black-hole/
People have pointed out the similarity to the mainframe era in a number of ways and this is one of them.
When the only way for consumers to get their own personal "world-changing" computation accomplished was by relying on expensive remote data centers (because that's where the mainframes were), it was never going to fly off the shelf. And there was plenty of time for it to mature as fully as it could.
Nothing is any real help to consumers or even most businesses until what are still known as "data centers" are completely un-necessary to them. Like when PCs came along and were powerful enough to finally exceed the abilities that centralized mainframes had to offer, and at a virtually insignificant cost by comparison. An unbeatable combination.
AI can't be expected to yield its best financial performance for consumers either, until everything the massive data centers being built have to offer, has been exceeded by a typical affordable desktop PC. Again. But this time in the 21st century.
Nothing less will do, until then might as well be still tied to a mainframe.
Which was not too bad IIRC, but limited to serving the very few, compared to PCs which ended up capable of serving almost anybody.
And mainframes are by no means completely obsolete, but gained widespread recognition as passe already decades ago,
It's not going to be as good as it could be, for investors especially, until expensive data centers are equally as passe.
Before PCs, the projected need for increasing tonnage of future mainframe data centers (which incidentally have somewhat long delivery times) was pretty reliable, until one day it wasn't. I wouldn't have wanted to be one who was still bound to take delivery after that.
By the same token indications are that today's new data center construction momentum will "eventually" outrun the need to complete every single project even after ground has been broken.
This could be big simply because, with the scale it is happening at, it cannot possibly be small :\