AI is putting an end to affordable smartphones. The memory that fuels your device is now being redirected to data centers instead.

AI is putting an end to affordable smartphones. The memory that fuels your device is now being redirected to data centers instead.

      **Summary**: Memory manufacturers have shifted their focus from producing chips for smartphones to those for AI, leading to a 250% increase in LPDDR prices and a 59% collapse in India’s sub-$100 phone market. In 1985, an IBM PC AT cost $19,400 in today's dollars, whereas today a Tecno Spark Go is just $30 yet far exceeds that PC in speed. The trend of declining costs is now ending, as the International Data Corporation forecasts a 13% drop in global smartphone shipments by 2026, with Africa and the Middle East seeing an even greater decline of over 20%. This downturn predominantly affects the low-end market, prompting IDC to label it "a structural reset."

      The primary driver of this shift is the demand for DRAM, which is hard and costly to produce. For years, the bulk of DRAM supply served smartphones and laptops, but in recent years, AI's rise as a significant consumer of memory has redirected production to AI data centers. Samsung, SK Hynix, and Micron account for over 90% of global DRAM output, and new fabrication plants require substantial investment and time to become viable. These manufacturers have historically kept supply tight to avoid past financial pitfalls.

      The emergence of high-bandwidth memory (HBM), essential for AI computing, has shifted priorities. HBM's production is growing, with its profit margins significantly exceeding those of standard DRAM. Instead of increasing overall production, manufacturers are reallocating current capacity to HBM, severely impacting the supply of memory for devices like phones and laptops. By the end of 2025, SK Hynix was directing 30% of its wafers towards HBM, while Micron discontinued its consumer brand entirely to focus on AI and enterprise needs.

      The repercussions have been severe for budget smartphone manufacturers like Transsion, Oppo, Vivo, and Lava. With rising memory costs, the viability of sub-$100 smartphones has plummeted, resulting in Transsion's profits dropping 54% in 2025 and significant reductions in their shipment targets. In India, sub-$100 smartphone sales have fallen by 59% year on year, and many consumers in Africa are being priced out of ownership altogether.

      While memory makers are profiting immensely—earning $70 billion collectively in 2025 and projected to more than double that in 2026—the fallout is impacting consumer sectors. Companies like Samsung and Dell have raised prices significantly, and Apple has faced challenges in securing long-term memory agreements, leading to increased costs for components in its products. The trend shows no signs of slowing, with forecasts predicting that memory could represent 45% of the iPhone's component cost by 2027.

      China's ChangXin Memory Technologies is currently the only potential source for near-term relief, even as it pivots some of its capacity to more profitable HBM production. Hyperscalers like Microsoft and Google are outbidding budget phone makers for DRAM access, further complicating the landscape. This memory crisis reveals a growing divide as the wealth generated by AI consolidates amongst a few memory producers, while the poorest consumers bear the burden of increased prices and reduced access. The era of affordable computing appears to be over, with the most disadvantaged feeling its effects first.

Other articles

AI is putting an end to affordable smartphones. The memory that fuels your device is now being redirected to data centers instead.

LPDDR prices increased by 250% over the course of a year as Samsung, SK Hynix, and Micron shifted their wafers towards AI. The sub-$100 smartphone market in India plummeted by 59%.