Why the AI Memory Shortage Is Just Getting Started (and Who Wins From It)
Everyone is chasing AI chips, but the biggest investing opportunity may be hiding in the memory shortage powering the entire AI revolution.
Overview
Most AI conversations these days revolve around processors like Nvidia's. But a crisis is unfolding underneath the boom, and it may prove just as important for investors. The world is running short on memory chips, the components that store the data AI systems devour, and the shortage is not winding down. If anything, it is just getting started. Here is why, and who stands to win.
The heart of the problem is a special kind of memory called high-bandwidth memory, or HBM, which sits beside AI chips and feeds them data. Each new generation of AI hardware demands far more of it, and here is the crucial twist: Making HBM is enormously capacity-hungry. A gigabyte of HBM consumes roughly 3 times as much factory space as an equivalent amount of ordinary memory. So, as the three big manufacturers direct their production toward high-margin HBM, they are starving the supply of everything else.
Details
That is why analysts now expect memory prices to keep climbing into 2028, with one memory giant warning the crunch could stretch past 2030. New factories take years to build, and the capacity coming online will barely keep pace with demand. Simply put, the supply-demand gap is widening, not closing.
Source
Originally published at www.fool.com.