Have you shopped for a new laptop or smartphone lately? It’s getting rough out there.
A survey published earlier this year by Gartner projected average PC prices would rise by 17% this year, and average phone prices by 13%, with cheaper models paradoxically seeing the biggest price increases.
Why? Two words: memory crisis.
The price of computer memory chips (e.g., random-access memory, or RAM) is going through the roof because of skyrocketing demand from AI data centers. And while that’s bad for shoppers, it’s good for shareholders of certain memory chip stocks and ETFs.
Why does AI need so much memory, anyway?
So that it can remember things, duh
If you’ve been playing with AI for a while, you may remember that chatbots used to lack object permanence, like a newborn. Up until a couple of years ago, even the best AI models could remember, at best, the last few messages you’d sent them. They had zero recollection of previous conversations.
Now, it’s different. AI labs like Anthropic and OpenAI have realized that consumers want to be able to have long conversations and tackle complex projects with their AI models, without having the models lose coherence and forget what they’re doing midway through. And consumers want to be able to come back to old conversations and projects over multiple days.
That means that the models need to be able to “remember” things, to put it anthropomorphically. Or, more accurately, the models need to have much longer context windows.
Hence the RAM bottleneck. Memory chip makers are using up all their production capacity fulfilling huge orders of high-end chips for AI data centers, which is starving the consumer electronics supply chain (particularly the budget segment) of normal memory chips. And the bottleneck is only expected to get worse through the end of the decade.
Again, this is bad for anyone who wants a cheap phone or laptop. But it’s great for certain memory chip stocks and ETFs.
