> While most people blame massive demand from AI companies, Micron CEO Sanjay Mehrotra says price pressure from certain customers over the past few years has also significantly contributed to the shortage.

This is some next level gaslighting. AI companies significantly increased the demand and have deeper pockets than most consumers. He who bids most gets the product.

> Mehrotra also said the memory shortage could last into 2027 and beyond because fabrication plants take years to build, and the next generation of memory is even more complex to manufacture. Micron is now investing up to $200 billion in manufacturing and R&D, including new memory fabs in Boise, Idaho, and Syracuse, New York.

When the AI bubble pops, and supply significantly outpaces demand, we'll then see them close tonnes of fabs.

I don’t think it’ll necessarily pop. The excess supply could even induce additional demands in not just LLM or AI but in other verticals. The margins of hyperscalers and infra providers might thin out, but with all that surplus in hardware, they can find new uses. For example, developers rent their own GPU node for cheap capable of running frontier open weight models instead of running local models. Computers and consoles will be cheaper again and more people will want to buy them. Non-LLM related simulations could benefit from excess hardware.

"When the AI bubble pops"

Too many people seem to believe that the AI bubble popping means something like "the total volume of the activity will precipitously drop and whole centers will be abandoned forever".

I don't believe that. I was an adult when the dot-com bubble popped. What happened then was that a lot of unsound businesses went under, and the healthier players (like Amazon) expanded into the resulting void. Ultimately, the relentless march of digital technology didn't even slow down back then, although the inflow of risk capital definitely did. But the existing resources were mostly taken over by someone else.

If/when the current bubble bursts, it will be similar. All that infrastructure will be resold and re-rented to healthier players. By now, machine learning has reached a level of usability/maturity which is genuinely useful and the economy at large won't abandon it any more than it abandoned e-mail or WWW back then.