Skip to content
AI supply chain term

HBM (High-Bandwidth Memory)

High-Bandwidth Memory — DRAM stacked next to an accelerator to feed it data fast; often the real limit on GPU output.

What it means

High-Bandwidth Memory is DRAM that is stacked vertically and placed right next to an AI accelerator, then wired to it through thousands of connections. An AI chip can only compute as fast as it can be fed data, and ordinary memory cannot move bits quickly enough — so HBM exists to keep the most expensive silicon in the data center busy. Each new generation (HBM3, HBM3E and beyond) widens that data pipe. Because the stacks are hard to manufacture and must be co-packaged with the processor, HBM is frequently the part of the chain that actually caps how many accelerators the world can ship, rather than the logic chip itself.

Why it matters to investors

When HBM is the binding constraint, the handful of memory makers that can produce it hold real pricing power, while accelerator vendors that depend on it can be supply-limited no matter how strong demand is. HBM tightness ripples straight into how many GPUs ship, which is why it is one of the most watched chokepoints in the entire AI build-out.

See HBM in the AI value chainIts live model score, why it matters, and every company exposed to it.

Companies on this part of the chain

SK hynix000660.KSMicronMUSamsung Electronics005930.KSNVIDIANVDA

Named to show where the term sits in the AI supply chain — research, not advice, and never a recommendation to buy or sell.

Related terms

DRAM (Dynamic RAM)Dynamic RAM — the main working memory; HBM is a stacked, high-bandwidth form of it.CoWoS (Chip-on-Wafer-on-Substrate)Chip-on-Wafer-on-Substrate — the advanced packaging that fuses compute dies and memory into one accelerator.Advanced packagingAssembling compute dies and memory into a finished, working chip — a real capacity bottleneck.GPU / AI acceleratorGraphics Processing Unit — the massively-parallel chip that does most AI training and inference.Memory bandwidthHow fast data can move per second — often the binding limit, not raw compute.

See HBM in the live AI chain.

THE ENTITY maps every constraint onto one live model — which part is tight now, who owns it, and who gets squeezed when it moves. Plain-English reads you can check.

Explore the features
All glossary terms

THE ENTITY is an educational read on the AI supply chain — research, not investment advice. It explains how the chain works and who sits where, never price targets or buy/sell calls.