@ahmetcantryk
Short answer: phones and PCs are getting more expensive in 2026 mainly because AI data centers are buying up most of the world's DRAM and NAND memory production. Gartner projects a combined ~130% surge in memory prices by the end of 2026 versus 2025, pushing PC prices up 17% and smartphone prices up 13%.
Why did memory prices rise so much?
Demand for AI inference systems and hyperscale data centers keeps DRAM and NAND supply constrained, while memory manufacturers shift production capacity toward higher-margin server products. Analysts estimate AI data centers could consume roughly 70% of high-end DRAM in 2026 — a sharp inversion from prior cycles.
According to TrendForce, conventional DRAM contract prices rose 90–95% quarter-over-quarter in Q1 2026, slowing to a 13–18% increase by Q3 2026, while NAND flash contract prices rose 10–15% over the same period. Samsung, SK Hynix, and Micron have collectively shifted 93% of their combined production capacity toward high-bandwidth memory (HBM) for AI data centers, shrinking the share left for general-purpose RAM in consumer devices.
Why do AI phones and AI PCs need more memory?
On-device AI features — local summarization, real-time translation, image processing — need model parameters held in memory, pushing manufacturers to ship phones with 12–16 GB of RAM instead of the 6–8 GB that used to be standard. That extra memory means putting more of an already-more-expensive component into every device, so the AI features themselves and the hardware cost they carry both push the price up.
The same pressure hits PCs: the "AI PC" category, built to run local AI assistants, needs more RAM and faster storage than previous generations, which also gives manufacturers an easier case for justifying the price increase as a premium AI feature. The same logic explains why AI earbuds with live translation have stayed relatively expensive — the on-device speech model needs memory even in a tiny form factor.
How big is the price increase, and which products are hit hardest?
The table below summarizes different analyst firms' end-of-2026 projections:
Source | PC price increase | Phone price increase | Note |
|---|---|---|---|
Gartner | 17% | 13% | Versus 2025, based on a ~130% combined memory-price surge |
IDC (pessimistic scenario) | 10–20% | ~8% average, higher at the low end | Thin-margin entry-level devices hit hardest |
Gartner (shipment impact) | Shipments down 10.4% | Shipments down 8.4% | Higher prices are also suppressing demand |
Entry-level devices are disproportionately affected because margins in that segment are already thin, forcing manufacturers to pass the rising memory cost straight through to the sticker price.
How are manufacturers justifying the price hikes?
As memory costs rise, manufacturers have started framing device price increases not just as inflation but as "you're getting more AI capability this time." That framing is partly true — on-device models genuinely need more memory — but it also gives manufacturers an easier way to pass rising component costs on to customers.
This lines up with what an NPU actually does in an AI PC: the NPU supplies the processing power, but memory is where the model weights that power runs on actually live, and together they form the technical justification behind the price increase.
Should you buy now or wait?
The case for a near-term price drop is weak: research indicates elevated prices and tight allocation could persist through 2027 as new production capacity lags demand, and manufacturers like SK Hynix have warned the shortage could last past 2030. That means "prices will come down if I wait a few months" isn't a safe assumption for 2026.
The practical call breaks down like this: if your current device is genuinely failing or losing support soon, waiting has little upside — prices are more likely to keep climbing in the near term. If your device still does the job, holding out for a possible easing once manufacturers bring new HBM capacity online after 2027 is reasonable, but it's a possibility, not a guarantee.
What does this mean practically for buyers?
For budget-limited buyers, the least effective move is stepping down a RAM or storage tier to save money now — buying a low-RAM device while memory prices keep climbing can shorten how long that device stays usable. For a device meant to last several years, choosing a higher RAM or storage tier now, if the budget allows, reduces the risk of an early forced upgrade two years from now because the memory ran out. For anyone weighing a wearable purchase too, the 2026 AI smart glasses comparison shows a similar memory-versus-price trade-off.
Why aren't memory makers prioritizing consumer RAM?
High-bandwidth memory (HBM) sold into AI accelerator cards carries a far higher profit margin than standard consumer DRAM — for manufacturers like Samsung, SK Hynix, and Micron, that turns which product gets a production line into a simple economic calculation. If a line can produce either HBM or standard DRAM, the margin gap pushes the manufacturer toward HBM almost automatically.
That means consumer-electronics makers have lost negotiating leverage with memory suppliers: where a large phone maker could once pressure prices down through bulk orders, AI data center customers are now willing to buy larger volumes at higher prices, weakening the consumer side's bargaining position.
What pressure is this putting on the software side?
Hardware makers facing memory costs they can't fully pass through are holding RAM amounts steady on some devices and pushing optimization to software instead — shrinking models to use less memory on-device through techniques like quantization and distillation is a direct result of that pressure. That's actually good news for consumers in the short term: as hardware gets pricier, more efficient models that do the same job with less memory are improving at roughly the same pace.
That optimization has a ceiling, though — past a certain point, shrinking a model costs accuracy, so manufacturers aren't presenting software optimization as a full offset to rising memory costs, just a way to soften the impact.
Frequently Asked Questions
Why did phone and PC prices go up in 2026?
The main driver is AI data centers buying up most of the world's DRAM and NAND memory production capacity. Gartner projects a combined ~130% surge in memory prices by the end of 2026, pushing PC prices up 17% and phone prices up 13%.
When will memory prices come back down?
Research firms indicate elevated prices and tight supply could persist through at least 2027 as new production capacity comes online, with some manufacturers warning the shortage could last past 2030. A significant near-term price drop isn't a realistic expectation.
Do AI features really need more memory?
Yes — on-device AI models like local summarization and translation need their parameters held in memory, requiring more RAM than previous-generation devices. That technical justification also makes it easier for manufacturers to pass rising memory costs on to customers.
Is it better to buy a device now or wait?
Since prices are unlikely to drop significantly in the near term, waiting has little upside if your current device genuinely isn't meeting your needs. If you're planning for medium-term use, choosing a higher RAM or storage tier now, budget permitting, reduces the risk of an early forced upgrade.
Does the memory crunch hit every device category the same way?
No, entry-level phones and PCs are disproportionately affected because margins in that segment are already thin, forcing manufacturers to pass rising memory costs straight into the price. Premium devices have more margin to absorb the increase, so the price hike is felt less as a percentage.