Anonymous #1 19h

If AI completely halts, do you think the operators of data centers will continue to maintain them at a deficit until we reach a point where they’re once again profitable? They’ll close as many as possible because the higher retained earnings will look better to investors. This is entirely propped up by AI’s requirements.

5

Anonymous #2 18h

Well two points abt that. 1. It’s a huuuge assumption to say that ai is a bubble that will entirely pop. I think there’s essentially a 0% chance that we just “completely halt” ai and a very low chance that development slows significantly. Some silly use cases will fall by the wayside sure but I think they’ll be replaced by more and more profitable use cases as those things reveal themselves.

1

Anonymous #1 18h

And 2. Data centers aren’t only profitable when used to power ai, they’re profitable doing all sorts of other things. As humanity grows, in population and technological complexity, we’re just going to need more computing power like we’re going to need bigger and more robust energy grids. So many industries are chomping at the bit to get computing time to run all sorts of models. We’re still in the age of waiting days to weeks for things like climate change prediction models to run and deliver

1

Anonymous #1 18h

Results. There will always be demand for more computing power to enable things like that to be done quicker and more accurately

1

Anonymous #1 16h

Well, to talk about #1, when the bubble bursts, we can assume that most speculative and weaker AI companies will be wiped out while the strongest players like Nvidia, Google, and AMD will survive and grow long-term. This one is backed by the .com bubble and its outcome.

1

Anonymous #1 16h

As for #2, data centers are extremely important overall, I’d absolutely agree. However, with an AI bubble bursting, it’d be unprofitable to continue operating a large amount of these data centers for likely a decade or more. With an AI bubble bursting, the demand to store data would plummet, the supply of data to centers would plummet, and the technology still isn’t quite at a place to make up for the deficit in data

1

Anonymous #2 16h

But again, “when” is incredibly overconfident. “If” seems much more pragmatic.

1

Anonymous #2 16h

Demand won’t plummet at all, that’s my whole point. Demand for computational power and data storage is increasing rapidly and won’t stop increasing simply because ai falls by the wayside, if that even happens. There are countless non ai digital goods and services that rely on data centers and would pay lots to have access to them, not to mention demand from governments and researchers.

1