The Center for Strategic and International Studies (CSIS) has published an analysis titled "Beyond the Memory Cycle: AI, HBM, and the New Semiconductor Shortage," arguing that demand for high-bandwidth memory (HBM) tied to artificial intelligence workloads is producing a supply squeeze that no longer follows the traditional boom-and-bust rhythm of the DRAM industry.

The report's central claim, stated directly in its title, is that the industry has moved "beyond the memory cycle." In past decades, DRAM and NAND pricing swung through predictable multi-year cycles of capacity additions, oversupply, price collapse, and consolidation. CSIS contends that AI accelerator demand — and specifically the HBM stacked next to GPUs and other AI processors — has changed that dynamic.

What does CSIS say is different this time?

The analysis frames HBM as the key variable. The memory type, which stacks DRAM dies and places them close to the compute silicon through an interposer, has become a binding constraint on AI system shipments rather than a commodity input. That inversion, CSIS argues, is what turns a conventional memory upcycle into what the report calls a "new semiconductor shortage."

The piece positions this shift alongside the broader AI buildout. As hyperscalers and AI developers compete for advanced logic and the memory that feeds it, supply-chain tightness migrates from one chokepoint to another — and HBM capacity, because it is technically difficult to expand and qualified slowly, sits near the front of that queue.

Why does the framing matter?

CSIS is a Washington-based policy institution, and its decision to frame the memory market as a strategic shortage rather than a commercial cycle carries weight in policy circles. Reports of this kind feed into U.S. debates over export controls, fab subsidies, and how allied governments treat leading-edge memory capacity.

For procurement teams and chip buyers, the analytical point is straightforward: if CSIS is right that the cycle logic has broken, then expectations of a familiar price correction — the moment when new capacity floods the market and HBM pricing softens — may not arrive on the old schedule. Memory vendors, in that reading, retain pricing power longer than historical models would predict.

What the report does not settle

The CSIS piece is an analytical argument, not a capacity disclosure. It does not replace vendor guidance on HBM bit growth, wafer allocations, or pricing. Readers looking for confirmed numbers should track the memory makers' own quarterly disclosures and the capacity plans of their foundry and equipment suppliers, which remain the ground truth for how quickly HBM supply can actually scale.

The open question the report leaves for the industry is whether HBM-focused capacity additions will eventually re-create classic oversupply — or whether AI demand growth keeps absorbing incremental output, entrenching the shortage conditions CSIS describes.