The AI Memory Shortage Just Entered Year Two and These 3 ETFs Own Every Layer From DRAM to HBM - 24/7 Wall St.

The AI memory shortage that began in 2025 has now entered its second year, intensifying pressure on hyperscaler supply chains. The structural demand floor created by AI inference workloads has signifi... Read original →

The AI memory crunch’s second year reveals HBM’s deep dependency on advanced packaging and EUV lithography. Upstream, ASML and Lam Research face capacity strain from sub-3nm demand; downstream, AI accelerator makers scramble to pre-book SK Hynix and Micron’s HBM3e output, disrupting DRAM’s historic cyclicality. Geopolitical friction is inflating compliance costs—U.S. export controls have forced Samsung and SK Hynix to reconfigure their fabs in Xi’an and Wuxi, blunting supply agility. With HBM4 slated for 2027 ramp, TSMC’s CoWoS capacity emerges as the next chokepoint, likely prompting NVIDIA and AMD to secure allocation via equity ties. Over the next 12–24 months, memory will shift from a cost center to a performance determinant in AI systems, channeling ETF flows into SMH and DRAM, while firms lacking an HBM roadmap—like Nanya and Winbond—risk strategic irrelevance.