Chip Crunch: AI Boom Drives Surge in Memory Prices and Supply Risk
The accelerating artificial-intelligence build-out is creating unexpected ripple-effects in the semiconductor supply chain: memory chips traditionally considered “mundane” are now facing price spikes and shortages as manufacturers shift capacity toward high-bandwidth AI-specific chips.
DAte
Oct 20, 2025
Category
Technology & Semiconductors
Reading Time
5–6 Minutes
According to Reuters on October 20, 2025, the global rush to build AI-compute capacity — particularly high-bandwidth memory (HBM) for model training — is squeezing the supply of more “ordinary” memory types used in servers, PCs and smartphones. This unexpected surge in demand is pushing customers into panic buying and driving up prices across the board.
Memory manufacturers such as Samsung Electronics and SK Hynix, which dominate the DRAM market (≈70% combined share), are reallocating production toward HBM and other AI-oriented chips, reducing availability of standard DRAM and NAND supply for other markets. Analysts call this shift the onset of a “supercycle” in memory-chip pricing—one driven not by smartphones or PCs, but by the AI build-out and upgrade cycle in data centers. “There is definitely scrambling going on … double/triple ordering,” notes an industry executive.
This scenario poses several challenges:
OEMs and device manufacturers face rising input costs and supply uncertainty.
Data-center operators may find it harder to scale infrastructure affordably.
Smaller chip and memory vendors may benefit from tighter supply among the big players.
Key Highlights
AI infrastructure demand is tightening supply of standard memory chips, not just AI-specialized ones.
Leading memory suppliers are reallocating capacity toward high-bandwidth AI memory.
Industry sources describe current environment as a “super-cycle” in memory pricing triggered by the AI boom.
Markets relying on standard memory (PCs, smartphones, servers) may face cost/availability headwinds.
Why This Matters
Hidden infrastructure cost rise: As AI build-out accelerates, downstream consumers (PCs, servers) may face margin pressure due to rising memory costs.
Supply-chain diversion: Dominance of HBM and AI-specific memory may crowd out production for traditional markets, forcing diversification or new entrants.
Investment signal: A super-cycle in memory suggests that AI growth is not just centered on chips and software—every layer from power to memory to cooling is in flux.
Competitive advantage shift: Memory suppliers and device makers who anticipated this shift earlier may gain while those left behind struggle to catch up.
Source
Reuters – Full Article
Author
