What the AI Memory Shortage Means
The AI memory shortage is a market squeeze where explosive demand from AI data centers for NAND and DRAM absorbs most available supply, leaving fewer and costlier memory components for mainstream PCs and consumer devices. At the heart of the problem is agentic AI, which runs continuous, multi-step tasks and keeps huge datasets in memory. These systems demand Petabytes of storage and vast pools of DRAM to keep models and user data active. That demand has triggered a NAND price surge as suppliers prioritize high-margin AI data center demand over lower-priced PC contracts. As memory cost inflation spreads through the supply chain, PC component availability tightens, forcing manufacturers to ration configurations, sign long-term supply deals, and pass higher costs to buyers. The result is a structural tug-of-war between cloud-scale AI projects and everyday computer users.
NAND Revenue Hits a Record While PCs Stall
Agentic AI has turned NAND into one of the hottest components in tech. Counterpoint Research data shows NAND revenue surging 3.5 times year-on-year to USD 46 billion (approx. RM212.8 billion) in a single quarter, propelled by AI data center demand. Enterprise buyers already account for 43% of the NAND market and are projected to pass 60% by the end of this year, underscoring how AI data center demand now dominates supply. Meanwhile, the broader PC market is shrinking. IDC reports an 11.3% decline in global PC shipments, with volumes expected to fall to around 260 million units and stay weak until at least the end of this decade. This imbalance means memory makers enjoy soaring revenue, while PC vendors struggle with higher component costs and softer demand, a classic case of one sector’s boom fueling another’s slowdown.
How Agentic AI Starves the PC Market of Memory
Agentic AI workloads differ from earlier models because they chain together long-lived tasks, often spanning search, generation, and automation. That behavior needs immense pools of fast storage and DRAM to keep models, context, and logs accessible. As hyperscalers and enterprises expand AI data centers with several Petabytes of NAND each, memory makers shift production toward high-density, high-margin server products. PC component availability suffers as a result, especially for higher-capacity SSDs and DRAM kits. IDC expects global PC shipments to show no real recovery until around 2030, partly because rising memory costs and shortages weigh on upgrade cycles. While new AI PC platforms such as Apple’s MacBook Neo and Intel Wildcat Lake designs emerge, they are being launched into a market where memory cost inflation limits aggressive pricing and constrains how many AI-ready configurations vendors can offer.
HP’s AI PC Push Meets a Memory Cost Wall
HP’s latest results show both the promise and the pain of the AI memory shortage. The company reported USD 14.4 billion (approx. RM66.6 billion) in quarterly revenue, up 9% year-over-year, boosted by strong AI PC demand and a continuing Windows 11 refresh. AI PCs already make up 44% of HP’s PC shipments, and the company expects them to reach 60–70% in fiscal 2027. Yet this AI push is colliding with rising memory and storage input costs that HP says will peak in the fourth quarter, dragging down Personal Systems operating margins before a hoped-for improvement in fiscal 2027. HP is responding with long-term supply agreements, tighter planning, strategic inventory, and repricing. This mix of strong AI PC demand and memory cost inflation shows how vendors must carefully balance growth in AI-optimised PCs against the reality of constrained DRAM and NAND supply.
