AI Investing Glossary
What is HBM and why does it matter for AI?
HBM, high-bandwidth memory, is stacked memory bonded directly next to an AI processor so data can reach the compute fast enough to keep it busy. Every AI accelerator ships with it, each generation uses more of it, and only a few manufacturers can produce it at scale, making HBM one of the tightest bottlenecks in the AI supply chain.
Why it matters for AI investors
HBM turned memory from a pure commodity cycle into an AI supply story: capacity is booked far in advance and pricing power has shifted toward the manufacturers who can yield it. It is also a useful demand signal, since HBM orders track accelerator production directly.
Tracked stocks in this layer
The largest names in the Memory & Storage category, ranked by market cap:
Market data as of the close on . Prices are end of day, not live.
Related reading
- What are AI infrastructure stocks? The 12 layers, explained · AI infrastructure stocks are the companies paid to build AI compute: chips, memory, networking, data centers, cooling and power. The 12 layers and the catch.
Frequently asked questions
Who makes HBM?
Production at scale is concentrated in three memory manufacturers: SK Hynix, Samsung, and Micron. Qualification with the leading accelerator vendors matters as much as raw capacity, so share can shift between generations depending on whose parts qualify first, which is why HBM roadmaps feature heavily in all three companies' earnings calls.
Related terms
See how this fits the whole picture in the Start Here guide, or browse the full glossary.