AI data centers squeeze memory chips, pushing phone and PC prices up

Memory chips, which temporarily hold data in computing devices, come in two main types: DRAM, used in laptops and smartphones, and high-bandwidth memory (HBM), used in large AI data centers to hold model weights and other data for services like ChatGPT, Claude, and Gemini. Although they differ significantly, both rely on overlapping wafer-fabrication capacity at a small group of manufacturers, including Micron and SK Hynix. Critically, HBM requires more wafers than conventional DRAM to produce the same memory capacity.

Recently, manufacturers have been allocating more of their limited capacity to the higher-margin HBM chips needed by data centers, leaving less capacity for DRAM. The result is a supply-side pricing problem: memory chip prices are estimated to have roughly doubled in Q1 2026. This shortage has forced consumer electronics makers to raise prices across the board. Apple has increased prices on Mac, iPad, HomePod, AppleTV, and Vision Pro, with analyst predictions of price increases for the iPhone. CEO Tim Cook has called these increases "unavoidable." Meanwhile, Lenovo, Dell, and HP—companies operating at considerably lower margins than Apple—have also signaled substantial price increases. IDC projects that the average PC selling price will rise 18.3% in 2026 while global PC shipments fall 11.3% due to production constraints.

The dynamic at work is a spillover effect in which end customers of consumer electronics—everyday buyers of phones and laptops—are now effectively competing against giant AI data centers for the same scarce manufacturing capacity. Data centers, with higher willingness to pay, are winning this competition. From a pricing standpoint, this shifts the market mechanism: when supply is constrained rather than demand, price becomes a tool to allocate scarce capacity to customers who can afford to pay more, rather than a tool to attract additional buyers.

Apple's own AI strategy is amplifying this pressure. The company has adopted a strategy of running AI models directly on devices rather than exclusively in the cloud, which preserves user privacy but increases memory demands. Its most powerful on-device models require devices with at least 12 gigabytes of memory—more than many existing iPhone models provide. Apple is thus squeezed from both sides: chip shortages reduce available DRAM while its on-device AI roadmap increases the company's need for memory in premium devices. As long as AI data center investment continues to accelerate, the supply constraint is unlikely to ease, and consumers should expect prolonged price increases across consumer electronics throughout 2026 and beyond.

The current wave of price increases in consumer electronics reflects a fundamental shift in how manufacturing capacity is allocated. Memory chip producers face a choice between serving consumer electronics makers and AI data centers—and data centers are winning because they can pay higher prices. DRAM and high-bandwidth memory (HBM) rely on overlapping fabrication capacity at a small group of manufacturers such as Micron and SK Hynix. Since HBM requires more wafers than conventional DRAM to produce the same memory capacity, the reallocation creates a classic spillover effect: as data center demand surges, the cost of DRAM for consumer devices climbs sharply.

Apple's situation illustrates the dual squeeze. The company's strategy of running AI models directly on devices (rather than only in the cloud) increases memory demands on premium phones—its most powerful on-device models require at least 12 gigabytes of memory. At the same time, chip shortages reduce the overall DRAM available to the company. Other manufacturers such as Lenovo, Dell, and HP operate at considerably lower profit margins than Apple and have even less room to absorb higher component costs, yet all are signaling price increases. IDC's forecast of an 18.3% rise in average PC selling price paired with an 11.3% drop in shipments in 2026 suggests that consumers will face both higher prices and reduced availability.

The deeper lesson is that seemingly unrelated markets—consumer electronics and AI infrastructure—are now tightly connected through shared production bottlenecks. As long as AI data center investment continues to accelerate, that connection will persist, and consumer device prices will remain under upward pressure.