Memory chips have become costly, owing to the high demand and limited supply of high-bandwidth memory (HBM). As a result, AI infrastructure deployments have also become more costly — that is if companies can get their hands on memory in the first place. Back in October 2025, Nvidia (NVDA) CEO Jensen Huang traveled to South Korea to meet with executives and secure memory supply. When the world first woke up to memory as a bottleneck, Nvidia already had a head start.
Still, while Huang did something similar in June of this year, he hasn't been able to make the company immune to the memory supply shock. AI server racks are becoming hard to build with limited memory, and the problem is now finally knocking on Nvidia's door.
Nasdaq Futures Plunge as Samsung Sparks Chip Selloff
AbbVie vs Eli Lilly: 1 Is Clearly the Better Dividend Stock to Buy and Hold for the Next 10 Years
The Nasdaq-100 Could Be Forming a Textbook Diamond Top. Here's What to Watch on the QQQ Chart Now.