Why the iPhone Is About to Get More Expensive

The Verge describes the situation as “chipflation” or “RAMageddon,” saying it has reversed decades of falling memory costs that kept consumer electronics powerful without making them dramatically more expensive. AlphaSense data cited by The Verge shows that “memory prices” and “memory shortage” appeared in 473 company transcripts last quarter. The industry is working on the problem, but the report says it may take years for supply to catch up.

AI is widely blamed for increasing demand, but The Verge says the shortage was brewing well before ChatGPT took off. Generative AI amplified the crunch with an unending hunger for memory. Micron president and chief operating officer Manish Bhatia told The Verge: “We need to build more wafer capacity.” He described this as a very different challenge from earlier years, when technology alone could keep up with demand.

For years, memory makers could increase output by fitting more chips onto each wafer. Those gains were shrinking and slower to achieve. By 2021, Micron concluded that technological advances alone would no longer create enough capacity, and manufacturers would need to process more wafers in enormous new facilities. The market then collapsed after pandemic-era purchases pulled demand forward and consumers pulled back. When it began recovering, generative AI created demand far larger and more memory-intensive than manufacturers had anticipated.

Supply is extraordinarily concentrated. Counterpoint data cited by The Verge puts Samsung at 39 percent of the memory market in the second quarter of 2026, SK Hynix at 26 percent, and Micron at 25 percent, giving the three companies about 90 percent of the world’s supply. They must divide limited capacity between AI infrastructure and consumer devices.

Memory in devices broadly falls into DRAM, used for active tasks, and NAND flash, used for long-term storage. AI data centers rely heavily on high-bandwidth memory, a form of DRAM that can move data far faster. Counterpoint’s David Naranjo told The Verge: “It’s not as simple as saying data centers are consuming RAM. The RAM is not the same.”

HBM stacks chips and uses advanced packaging, making it difficult to produce but lucrative to sell. It also consumes far more manufacturing capacity. Micron estimates that producing a given amount of HBM requires about three times as many wafers as the same amount of conventional DRAM. Companies including Nvidia and Meta are willing to pay a premium, so memory makers increasingly favor AI buyers and can lock in multiyear commitments instead of forecasting consumer device sales. Jaejune Kim, quoted by The Verge, said: “We have been engaging in discussions with customers, prioritizing those who can guarantee committed future captive demand.”

Editor’s Summary: Apple’s new iPhones are likely to carry higher prices as memory costs surge. The Verge found the shortage began before ChatGPT and is reinforced by AI’s demand for high-bandwidth memory, with limited supply expected to persist for years.