The smartphone market has looked unusually bleak in recent months, and one of the clearest signs is coming from the world’s top chip manufacturer. With demand for 4nm smartphone chips sliding fast, TSMC is reportedly shifting increasingly underused 4nm capacity toward still-tight 3nm production, effectively rebalancing its factories around the parts of the market that remain strongest.
Reports from Taiwan indicate that TSMC has already started converting some 4nm production lines to support 3nm output, driven largely by the fact that non-Apple smartphone customers are revising their 4nm demand forecasts downward. In other words, the usual wave of orders for chips that power many mid-range and budget phones isn’t arriving like it used to, leaving portions of 4nm capacity sitting idle.
This shift isn’t as simple as flipping a switch. While a large portion of the equipment used for 4nm manufacturing can be reused for 3nm production—estimates place that overlap around 80% to 90%—the transition process is still complex and time-consuming. TSMC is said to expect the conversion to take roughly 6 to 12 months, reflecting the practical challenges of adapting advanced semiconductor lines.
The pressure is also visible upstream in chip ordering behavior. MediaTek and Qualcomm, two of the most important suppliers for Android phones worldwide, are believed to have reduced their 4nm production pace. Since 4nm silicon is heavily used across low- and mid-tier smartphones, cuts at this level ripple quickly through the supply chain. The reported reduction equates to roughly 20,000 to 30,000 wafers—translating into about 15 million to 20 million mobile chips that would otherwise have been produced.
A major force behind this slowdown is not just weaker consumer demand, but a sharp squeeze on smartphone component economics—especially memory. Even if you’ve seen headlines about spot DRAM prices easing, the broader pricing picture for mobile memory remains punishing. LPDDR5 contract pricing has surged dramatically since early 2025, rising roughly threefold and hovering around $10 per gigabyte. And forecasts point to additional double-digit percentage increases into 2027, keeping costs elevated for manufacturers that rely on large memory configurations to compete.
These memory costs hit the entry-level and mid-range smartphone segments the hardest because those devices have the least pricing flexibility. When a phone is designed to sell on value, there’s limited room to raise prices without losing buyers. Yet memory is taking a bigger and bigger slice of the bill of materials. DRAM alone is now estimated to represent around 35% of the BOM for an entry-level handset, while NAND contributes another 19%. Combined, that’s about 54% of a budget smartphone’s cost tied up in just two components—DRAM and storage. As memory becomes a dominant cost driver, it becomes harder for brands to build attractive low-cost phones, which in turn dampens orders for the kinds of chips most commonly produced on mature nodes like 4nm.
A recent comment from Xiaomi President Lu Weibing captures how extreme the situation has become for phone makers trying to protect “value for money” pricing. He said the scale of memory price increases has exceeded expectations, claiming that the same memory configuration has jumped nearly fourfold compared with the first quarter of last year. He added that a 12GB+512GB model rose by around RMB 1,500, while a 16GB+1TB version increased even more dramatically—changes that put serious pressure on value-focused product lines.
Put together, these trends help explain why TSMC is reallocating resources. If low- and mid-tier smartphones are getting squeezed by soaring memory costs and weakening demand, foundries will naturally pivot capacity toward more advanced processes and customers that are still placing large, stable orders. The end result is a chipmaking landscape where 3nm capacity remains in high demand, while 4nm production for mainstream smartphones increasingly has to fight for its place.
