One of the most tangible consequences of the AI boom is arriving somewhere unexpected: the bill of materials inside phones, laptops and other ordinary electronics. The same manufacturers serving consumer devices can earn far more by directing capacity toward high-bandwidth memory for AI data centers.
Samsung, SK Hynix and Micron control roughly 90% of the memory market. AI infrastructure uses enormous amounts of high-bandwidth memory, or HBM, which is harder to manufacture and more profitable than conventional DRAM used in consumer devices. That creates a simple economic pull: scarce factory capacity moves toward the highest-value customer.
Some memory components have reportedly quadrupled in price. New factories are being built, but semiconductor capacity does not appear overnight; projects can take years before meaningful volume reaches the market. That makes this less like a temporary holiday shortage and more like a supply problem device makers have to design around.
For buyers, there are several ways the pressure can show up without a manufacturer ever saying 'AI surcharge.' A laptop can keep the same price but ship with less memory. A higher-memory configuration can become disproportionately expensive. SSD pricing can stop following the familiar pattern of getting cheaper every generation. Or the entire device can simply move up a price tier.
For D, the useful behavior is not panic-buying. It is paying more attention to memory and storage when a purchase is already planned. Those are increasingly the components most exposed to the same supply chain feeding AI infrastructure, and they are also the parts Apple and many thin laptops make difficult or impossible to upgrade later.
The bigger point is that the AI buildout is beginning to have second-order effects outside AI products themselves. Cloud demand is competing for physical manufacturing capacity, electricity, networking hardware and now memory. That makes ordinary consumer electronics one of the places to watch for the hidden cost of the boom.