What it means
DRAM (dynamic random-access memory) is the main working memory of any computer, storing the data a processor is actively using; it must be refreshed constantly, which is why it is called dynamic. Server DRAM refers to the DDR modules, now generations such as DDR5, that populate the memory slots of data-center machines. In AI servers this memory holds datasets, model weights being staged, and the software environment around the accelerators, while also serving the host CPU. Crucially, server DRAM shares wafer capacity and process technology with high-bandwidth memory (HBM); when memory makers shift lines toward more lucrative HBM, conventional DDR can tighten. That makes server DRAM both a broad enabler of compute and a capacity that competes directly with the AI memory premium.
Why it matters to investors
Server DRAM is a cyclical commodity whose pricing is set by the same handful of memory makers that produce HBM, so HBM demand can pull DDR supply and prices along with it. A small group of large DRAM producers holds most of the pricing power.
Companies on this part of the chain
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
See Server DRAM 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.
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.