Phones, laptops get costlier as memory prices rise; Sam Altman warns buyers

DRAM and storage are two different kinds of memory inside modern devices. DRAM is short-term working space. It holds data that the processor is actively using, helping apps, games, and operating systems run. Storage chips keep files, apps, photos, and games for longer, even when the device is turned off. Phones, laptops, and consoles rely on both.

The distinction matters because each device needs a stock of these chips. A laptop may need DRAM for multitasking and storage for its operating system and files. A game console needs both to run games and save them. When chip supplies tighten, manufacturers face higher costs before products reach stores.

The article groups DRAM and storage together as “memory” and says prices have climbed steeply. Brands including OnePlus, Samsung, and Oppo passed some costs to buyers. It does not separate the shortage’s effect on DRAM from storage, but it shows that both are central parts of the gadget supply chain.

The article does not give one universal percentage for every memory chip. It says mobile-grade memory costs several times what it did a year earlier, while laptop memory rose dramatically as a share of manufacturing cost. That makes memory a much larger part of the price manufacturers must recover.

HP provided the clearest example. In July, memory represented about 35% of a laptop’s production cost. A quarter earlier, it represented only 15–18%. This means memory’s cost share more than doubled from the upper end of that earlier range, leaving less room for other components, assembly, and profit.

The increase helps explain why laptop prices and forecasts are under pressure. Intel expects the PC market to shrink 12% this year and another 10% in 2027. The company also does not expect memory supply or pricing to ease next year, so the cost burden may continue rather than quickly reversing.

When memory becomes scarce, manufacturers must pay more for a component every gadget needs. They can absorb that increase, reduce margins, change specifications, or pass the cost to customers. The article shows brands mostly passing it on, making phones, laptops, and consoles noticeably costlier.

In India, OnePlus, Samsung, Oppo, and others revised prices more than once. Phones below Rs 10,000 are especially at risk because even a relatively small component increase can make budget products harder to price. Sony raised PS5 prices twice this year, while the Xbox Series X is 60% more expensive in the United States than at its 2020 debut. Nintendo also raised the Switch 2’s price.

Higher prices can reduce sales because shoppers delay purchases or choose cheaper products. Intel expects the PC market to shrink 12% this year and another 10% in 2027. The article also reports Intel shares fell over 20% in extended trading after a weak forecast, showing concern about demand and costs.

AI data centres need large amounts of memory to run models and serve many users. Personal devices also contain DRAM and storage, so both groups draw from the same manufacturing capacity. When AI companies expand their computing infrastructure, they increase demand for chips that phones, laptops, and consoles already need.

The article presents Axiom as part of that competition. Sam Altman said Axiom has built its own chip, is learning to build data centres, and plans to use its robotics programme to help make more of them. He argued that moving open-source AI models onto personal computers would not necessarily solve the global shortage.

Instead, Altman said personal computers could hold large amounts of limited DRAM that are used only about 5% of the time. Data-centre memory can serve many users throughout the day. The result is a conflict between local ownership and shared capacity while memory supply remains constrained and new factories take time to build.

Altman’s argument is about utilisation, or how much of a resource is actively used. A laptop or phone may contain substantial DRAM even when its owner is browsing lightly, asleep, or not using the device. That memory remains unavailable to other people during those idle periods.

A shared cloud data centre can allocate memory across many users. When one user is inactive, capacity can support other workloads. Altman said this makes cloud memory more efficient in a supply-constrained environment, because data-centre resources can be used throughout the day rather than tied to one person’s device.

He rejected the idea that running open-source AI models on everyone’s personal computer would automatically reduce pressure on data centres. From a global allocation perspective, he called that approach “quite terrible” because it would place limited DRAM in machines where it may be used only 5% of the time. His solution is higher utilisation, alongside much more supply.

Memory manufacturers are benefiting from a market where demand exceeds available supply. Prices have risen, and customers are signing long-term agreements to secure future production. That combination creates unusually strong revenue and cash generation for chipmakers, even as device companies and buyers face higher costs.

SK Hynix reported a record quarterly operating profit of 83.3 trillion won and said more than half of its 2027 and 2028 production was already booked. Samsung’s preliminary July-to-September operating profit was around 107.4 trillion won, its fourth straight record quarter. These figures show how strongly the shortage is benefiting suppliers.

Yet manufacturers remember earlier memory cycles. Altman said they once spent $20 billion on capacity to earn $19 billion and nearly went bankrupt after building too much. That history encourages caution about new factories. The result is a difficult balance: profits are high now, but cautious investment means relief may not arrive quickly, and SK Hynix warned the shortage could last through 2030.

Supply and demand set the direction of memory prices. AI data centres, phones, laptops, and consoles all need chips, but factory output cannot immediately expand when demand jumps. Buyers then compete for limited production, pushing prices higher. Semiconductor factories require major investment, specialised equipment, and long planning and construction periods, so supply responds slowly.

The article gives a clear example of demand being committed ahead of time. SK Hynix said more than half its 2027 and 2028 production was already booked under long-term deals. Those agreements help customers secure chips, but they also show how tight future supply is. Memory makers are earning record profits while manufacturers pass costs to consumers.

Past overbuilding shapes today’s caution. Altman said memory companies once spent $20 billion on capacity to earn $19 billion and nearly went bankrupt. They therefore hesitate to add factories too quickly. With investment restrained and demand strong, the shortage may last through 2030, while gadget prices remain under pressure.