What the AI PC Memory Shortage Really Means
The AI PC memory shortage is a growing supply imbalance where AI servers consume NAND flash and DRAM faster than manufacturers can produce them, leaving fewer chips for consumer PCs and pushing up prices across the entire computer market. At the center of the crunch is NAND flash, the storage technology in SSDs and many devices. Counterpoint Research data shows NAND revenue has rocketed 3.5x in a single quarter to USD 46 billion (approx. RM212.7 billion), powered by data centers racing to support Agentic AI workloads. Enterprise buyers, building racks of AI servers packed with high-capacity SSDs, now account for 43% of NAND demand and are on track to exceed 60%. This shift pulls memory chip supply away from mainstream laptops and desktops, setting up a tug-of-war between cloud-scale AI and everyday PC buyers.
AI Servers Soak Up NAND as PC Demand Slips
NAND suppliers are enjoying a windfall even as PC makers struggle. According to Counterpoint Research, NAND revenue hit USD 46 billion (approx. RM212.7 billion) in Q1 alone, surpassing the entire NAND market revenue for 2023. The driver is AI server demand: Agentic AI models thrive on vast datasets stored on SSDs, so data centers are buying Petabytes at a time. Enterprise already represents 43% of NAND consumption and could top 60% by year-end, giving cloud and AI buyers priority access to the tightest part of the memory chip supply chain. The result is a starved PC market. IDC now expects global PC shipments to fall by 11.3% to about 260 million units, with meaningful recovery delayed until around 2028–2030, in part because higher NAND flash prices make new systems more expensive for both manufacturers and consumers.
HP’s AI PC Surge Meets Rising Memory Costs
HP’s latest results show how AI PC demand can boom even as memory pressure rises. The company reported USD 14.4 billion (approx. RM66.6 billion) in quarterly revenue, up 9% year-on-year, fuelled by AI-optimised PCs and a Windows 11 upgrade cycle. AI PCs already represent 44% of HP’s shipments and are forecast to reach 60–70% by fiscal 2027. Yet this success is colliding with climbing memory and storage costs. HP says input costs rose in its second quarter, will rise again in the third, and are expected to peak in the fourth, when Personal Systems margins should hit a low point before improving. CFO Karen Parkhill highlighted a plan built on long-term supply agreements, sharper planning, strategic inventory, and selective repricing, but the direction is clear: AI server demand is bidding up the same chips HP needs to keep AI PC pricing attractive.
How Memory Inflation Reaches Your Next Laptop
As AI server demand concentrates buying power in data centers, memory chip supply for consumer devices tightens and cost increases spill into retail shelves. NAND flash prices are rising on the back of record enterprise orders, and PC makers must either absorb those costs or pass them on through higher prices or reduced specifications. HP’s stronger Personal Systems margins this quarter rested on a “richer product mix” and disciplined pricing, but the company openly expects a margin trough as memory costs peak. Other OEMs face the same squeeze: they need more memory per system to support on-device AI features, yet every extra gigabyte is costlier. That dynamic threatens to turn AI PCs into premium-only products while budget buyers hold onto older machines longer, deepening the slump in traditional PC shipments even as AI-focused models gain share.
Long-Term Fix: More Fabs, But a Long Wait
Chipmakers are racing to expand memory chip supply, but relief for AI PCs and laptops is several years away. New DRAM and NAND fabs are planned across multiple regions, and companies like YMTC and CXMT are in what analysts describe as an epic expansion phase, aiming to roughly double wafer output. Meanwhile, YMTC has already grabbed a 13% share of the NAND market, growing revenue about 246% year-on-year, adding more competition but not yet enough capacity to cool prices. IDC expects global PC shipments to stay weak in 2026 and 2027, with a more solid recovery only around 2028–2030, by which time these new fabs should be online. Until then, AI server demand will keep dominating memory chip supply, and buyers of AI PCs and mainstream laptops alike should expect higher prices or leaner configurations.
