Jensen Huang Thinks the Artificial Intelligence (AI) Memory Boom Is Impossible to Ignore. Here’s My Top Pick That No One Is Talking About.
https://www.aol.com/articles/jensen-huang-thinks-artificial-intelligence-223500475.html
Publish Date: 2026-06-10 18:56:00
Source Domain: www.aol.com
Key Points
-
AI hyperscalers are increasing their appetite for high-bandwidth memory, DRAM, and NAND chips.
-
The AI memory supercycle has propelled Micron Technology, Samsung, and SK Hynix into the trillion-dollar club.
-
The Roundhill Memory ETF (DRAM) is an effective way to invest in the AI memory theme at a low cost.
During a recent trip to South Korea, Nvidia CEO Jensen Huang leaned into a key urgency rippling through semiconductor supply chains. Huang said, “The whole industry supply chain — everything from wafers to packaging to silicon photonics…everything’s in short supply because the demand is so high. It is going to persist for several years.”
I think this statement is a sobering assessment of a new reality. Artificial intelligence (AI) has turned memory into one of the most critical — and constrained — resources in hyperscale chip stacks. What was once a cyclical commodity has become a secular growth engine virtually overnight. This shift is important because, as Huang notes, it is not a temporary spike. Rather, insatiable demand for memory is becoming a multiyear reordering of supply and demand that will determine how fast AI scales.
Will AI create the world’s first trillionaire? Our team just released a report on the one little-known company, called an “Indispensable Monopoly” providing the critical technology Nvidia and Intel both need. Continue »
While most investors obsess over Micron and Sandisk, I think one of the best ways for investors to participate in the entire AI memory supercycle is through the Roundhill Memory ETF (NYSEMKT: DRAM), a specific fund built precisely for this moment.
A stock chart moving up and to the right over an electronic display of stock tickers.
Image source: Getty Images.
Why is demand for memory exploding?
Training generative models and inference deployments require enormous bandwidth between processors and memory. High-bandwidth memory (HBM), which is an advanced form of dynamic random-access memory (DRAM) layered with…
