Evaluating Memory Hog Cycles in the AI Era

Key Takeaways

• AI accelerators consume ~70% high‑end DRAM now
• PC/phone shipments projected to fall 11% in 2026
• Memory prices doubled in one quarter, hitting 25‑30% BOM
• Past cycles had cloud/crypto fallback; AI may replace
• Micron dropped consumer line, chasing high‑margin AI memory

Summary

The article revisits the classic memory‑hog cycle—where surging demand fuels price spikes, capacity expands, and oversupply triggers crashes—through the lens of today’s AI boom. It highlights that AI accelerators now consume roughly 70% of high‑end DRAM, pushing memory prices up nearly double in a single quarter and inflating memory’s share of PC/phone bills to 25‑30%. At the same time, IDC forecasts an 11% drop in global PC shipments in 2026, a traditional warning sign for memory investors. The piece argues that AI‑driven infrastructure could provide a structural floor that softens the usual bust, but risks remain if that demand falters.

Pulse Analysis

Memory cycles have long mirrored the agricultural hog model: a surge in demand drives prices up, manufacturers pour capital into new fabs, and by the time capacity arrives the market cools, flooding inventory and collapsing prices. The 2019 cloud‑burst and the 2022 Covid‑era bust illustrate how quickly oversupply can erode margins, leaving DRAM and NAND producers scrambling. Those historical patterns set the backdrop for today’s debate, where AI’s voracious appetite for memory could rewrite the script.

Currently, AI accelerators dominate roughly 70% of high‑end DRAM production, and memory prices have nearly doubled in a single quarter, pushing memory costs to 25‑30% of a typical PC or smartphone bill of materials. Simultaneously, IDC predicts an 11% decline in global PC shipments for 2026, a classic “hog‑cycle” warning sign. However, AI introduces four structural differences—higher density wafers, improved allocation efficiency, longer‑term contracts, and a shift toward memory‑centric workloads—that may act as a floor, absorbing excess capacity that would otherwise depress the market.

For investors and memory manufacturers, the key question is whether AI demand can sustain the inflated pricing long enough to offset the consumer slowdown. Companies like Micron are already pivoting, exiting low‑margin consumer lines to focus on high‑margin AI‑optimized memory such as HBM. While this strategic shift could preserve profitability, over‑building capacity without a reliable fallback could still trigger a bust. Stakeholders should monitor AI infrastructure spend, contract stickiness, and the pace of fab roll‑outs to gauge whether the 2026 pullback becomes a buying opportunity or the prelude to another steep correction.