Buy a $600 laptop in 2026 and, at a glance, it’ll look identical to last year’s model. Open it up, though, and there’s often a less pleasant surprise waiting: a dimmer screen and 8GB of RAM where you’d have gotten 16GB at the same price not long ago. This isn’t a one-off. It’s not an accident either.
In the first quarter of 2026, contract prices for conventional DRAM memory rose between 90% and 95% compared with the last quarter of 2025, according to data compiled by market research firm TrendForce, which forecasts another 58% to 63% increase for the second quarter. Translated into actual products, that means average smartphone selling prices could rise by 6.9% over the course of 2026, while LPDDR5X memory, used in both premium smartphones and some AI servers, could end up costing 78% to 83% more quarter over quarter.
There’s one cause, and it’s no mystery to anyone following the industry: three companies, Samsung, SK Hynix and Micron, control around 90% of global DRAM production, and over the past two years they’ve redirected a growing share of that output toward the high-capacity memory chips AI data centres demand. By the most recent estimates, those data centres will absorb roughly 70% of the world’s entire memory output in 2026, up from a share that sat around 20-30% as recently as 2022. What’s left for the consumer market has simply shrunk, and prices have adjusted accordingly.
The knock-on effect reaches anything built around a memory chip: graphics cards, SSDs, prebuilt PCs, smartphones, gaming consoles. Some manufacturers are responding quietly, trimming the default RAM in entry-level models while holding the sticker price steady; others are starting to warn customers openly that price increases are coming. For people in the industry, the current squeeze, painful as it already is, isn’t really the main issue. Duration is. New memory fabrication plants take years to build and ramp up to full output, meaning a structural mismatch between AI demand and industrial supply isn’t likely to resolve itself in a matter of months.
What’s still unclear is who actually ends up footing the bill. The biggest tech companies can absorb some of the higher costs to secure the supply they need to build their own AI models. Consumers buying a laptop or a console in the coming months will have a much harder time doing the same.
