Memory manufacturers have a fixed number of wafers they can process at any given time. That capacity must be split between three types of memory: DDR, used in desktops and servers; LPDDR, used in mobile phones and low-power devices; and HBM, used with GPUs for AI workloads.
Until recently, HBM received only 2% of wafer allocation. But the explosive growth of AI data centers is expected to push that figure to 20% by the end of 2026. A single gigabyte of HBM consumes more than three times the wafer capacity that a gigabyte of DDR or LPDDR does.
Memory Manufacturers Face Capacity Constraints
Only three large memory companies remain in the market: Samsung, SK Hynix, and Micron. These firms operate enormous fabrication plants that convert silicon wafers into memory chips. Building new fabrication capacity is extremely expensive and takes years, so the industry has learned from past boom-bust cycles to under-provide rather than overbuild.
The result is a structural shortage. With profit margins from HBM soaring, manufacturers are prioritizing its production. That leaves less wafer capacity for consumer-grade DDR and LPDDR memory. As those chips become scarce, their prices rise.
The Impact on Consumer Electronics
This repricing has already begun to affect consumer devices. Everything from laptops and tablets to smart TVs and gaming consoles relies on affordable memory. When memory costs go up, either manufacturers absorb the hit or they pass it on to buyers.
David Oks provided one of the clearest explanations of this trend, noting that the shift in wafer allocation is not temporary. The memory industry has learned from the extinction of its rivals that it is safer to under-allocate production capacity. Without new fabs coming online in time, the squeeze on consumer memory will persist for several years.
Stay ahead of the AI curve
The most important updates, news, and content — delivered weekly.
No spam. Unsubscribe anytime.
Sub-$100 Smartphones Feel the Pinch
The most visible impact so far has been in the low-cost smartphone market. Devices priced under $100 are particularly important in emerging markets across Africa and South Asia. These phones use LPDDR memory, which now competes for wafer space with the far more profitable HBM.
As LPDDR prices climb, manufacturers are forced to either reduce profit margins or raise prices on the cheapest phones. If the trend continues, consumers in price-sensitive regions may find that the affordable smartphone they rely on becomes noticeably more expensive.
The original piece by Oks carried a headline that called the phenomenon "AI is killing the cheap smartphone," though a rephrased version better captures the broader story: the memory shortage driving up prices across consumer electronics.
What This Means for the Coming Years
There is no quick fix. Even if memory companies wanted to increase wafer capacity, it takes years to build new fabrication plants. In the meantime, the demand for HBM from AI companies shows no signs of slowing. That leaves consumer memory production at a disadvantage.
For anyone buying electronics in the next few years, higher prices for devices that contain memory chips should be expected. The days of ever-cheaper gadgets may be on hold while the industry sorts out how to balance the needs of AI data centers with those of everyday consumers.
Related on Neura Market: