The AI trade has left the hyperscalers in the dust. What will it take for that to change?

AI expansion is colliding with a hardware bottleneck, leaving major hyperscalers—Amazon, Alphabet, Microsoft, and Meta Platforms—“behind” amid chip shortages that they cannot easily fix. The issue centers on high-bandwidth memory (HBM), a specialized DRAM type required for AI computing, where SK Hynix holds roughly 60% share, followed by Samsung and Micron at about 20% each. The tightness has pushed memory chip prices higher and prompted Apple to disclose price increases linked to makers shifting capacity toward HBM. Meanwhile, a broad memory-stock basket surged 41% over the past month, while the companies’ stocks fell as the Nasdaq rose nearly 1%. Microsoft and Meta cited higher component pricing when reporting large capex. Only Meta’s consumer ad exposure limits perceived returns on AI spending, and the supply-chain bottleneck extends to capital equipment firms like Applied Materials, Lam Research, and KLA.