The global race for AI infrastructure capacity is creating a tightening market for various memory chip segments, with supply-demand gaps widening due to production shifts.
In the context of major international manufacturers like Samsung, Kioxia, and Micron collectively exiting the MLC (Multi-Level Cell) NAND flash market, this once-unremarkable "veteran" product has seen its prices skyrocket by approximately 300% recently. The resulting supply shortage is posing potential impacts on industries such as automotive electronics, industrial control, and medical equipment.
The development pattern of the NAND industry involves capacity expansion through vertical stacking. MLC, as its name suggests, typically uses a two-dimensional planar stacking method, known as 2D NAND. The industry mainstream has now iterated to the TLC (Triple-Level Cell) and QLC (Quad-Level Cell) stages, entering the realm of 3D NAND.
Consequently, it is evident that upstream memory suppliers' decision to gradually phase out MLC production is driven by considerations of product iteration and the desire to allocate more capacity to higher-margin products. As overseas giants halt production, this also presents certain development opportunities for domestic supply chain companies.
Price Hikes and Supply Disruption Upstream memory manufacturers are exiting low-margin markets, including MLC NAND, under conditions of overall limited production capacity.
Recently, Kioxia announced the discontinuation of some older-generation products, including small-capacity TSOP packages, certain traditional Floating Gate 2D NAND, and third-generation BiCS FLASH products, which will gradually exit the market.
TSOP packaging is primarily used for low-capacity MLC NAND Flash memory. Some views suggest that Kioxia's announcement signals its potential exit from this market. In response, Seiji Ogawa, Chairman and President of Kioxia China, stated that this move is based on considerations of new technology replacing the old, a common phenomenon in the industry.
It's not just Kioxia. Previously, Samsung, SK Hynix, and Micron have also announced that MLC-related products are entering end-of-life phases, with no further expansion of related production lines.
According to research by TrendForce, SK Hynix and Micron have essentially limited MLC production to levels meeting existing customer demand, lacking motivation for expansion. Therefore, global MLC NAND flash production capacity in 2026 is projected to decline by 41.7% year-over-year, further exacerbating the supply-demand imbalance.
An industry professional noted that it is a normal industry practice for upstream memory manufacturers to phase out older-generation products driven by technology iteration and profit targets, based on lifecycle management needs. Simultaneously, the trend is to concentrate high-quality capacity on high-value-added segments like enterprise storage.
In this context of extremely limited capacity, the prices of "older" products like MLC NAND have begun to surge significantly.
An analyst from CFM Flash Market noted that supply of consumer-grade MLC eMMC from original manufacturers has basically ceased. Samsung's supply of automotive-grade MLC will stop in the first quarter of next year, leading to sharp price increases for MLC eMMC in the spot market, with extreme prices reaching up to $110 for 8GB eMMC.
A senior analyst from Counterpoint Research pointed out that according to their Monthly Memory Tracker data, MLC NAND prices have risen approximately 300% year-over-year. Taking 16Gb MLC NAND as an example, the price has increased from around $3.2 in Q1 2025 to a projected $12.9 in Q2 2026.
"This round of price increases is primarily driven by the supply-demand imbalance caused by the ongoing EOL (End-of-Life) of MLC products," the analyst further explained. The core reason for the rapid price surge is the significantly accelerated pace of the EOL process. "Historically, EOL typically occurred gradually over several years, but this round of MLC NAND phase-out has been compressed to about one year. Meanwhile, the rapid growth in AI demand is driving capacity allocation towards server-related products, further intensifying the MLC supply crunch."
End-market demand for MLC NAND primarily comes from industrial control, automotive electronics, medical equipment, and networking, where requirements for product reliability, write endurance, and long-term supply commitments are extremely stringent. The automotive industry is particularly affected.
The CFM analyst noted that automotive-grade MLC eMMC is mainly used in areas like T-BOX, instrument clusters, HUDs, and low-to-mid-range cabins. "Applications using MLC generally have low performance and capacity requirements, often not needing high-capacity UFS, where 'good enough' suffices. However, the verification cycle for automotive-grade chips and modules is very long, and the mass production lifecycle for automotive-grade components is at least 3-5 years. Automakers negotiate annual prices and supply with original manufacturers on a yearly basis, making them less sensitive to immediate changes in AI storage supply and demand. They previously underestimated the severity of the memory shortage and the manufacturers' determination to halt MLC lines to concentrate capacity on high-value server storage, making the automotive sector the most visibly impacted by the memory price hikes and shortages."
Domestic Substitution and Commercial Migration Facing the MLC supply gap, some capable manufacturers have begun to act.
TrendForce indicates that due to the supply chain disruption for long-term MLC NAND Flash supply, Macronix, which has long focused on the embedded and high-reliability memory market, has a relative advantage in meeting this niche demand. Macronix has reduced its original NOR Flash capacity to expand MLC NAND Flash supply.
The Counterpoint analyst pointed out that the current 2D NAND supply gap is difficult to fully bridge. "Among domestic manufacturers, Yangtze Memory Technologies Co. (YMTC) has the relatively largest available capacity. However, it will also prioritize meeting server-side demand for TLC and QLC products to capture the structural opportunities presented by AI, leaving limited room to alleviate the MLC shortage."
Naturally, end markets will eventually adopt newer-generation technologies and products. Therefore, transitioning to TLC products with more advanced process technologies is becoming a trend.
The CFM analyst stated that besides Kioxia continuing to supply automotive-grade eMMC until 2028, domestic memory manufacturers are also attempting to fill part of the supply gap through their own design and tape-outs. Meanwhile, the automotive industry will accelerate the verification and adoption of 64GB TLC eMMC/UFS.
"The downstream application of MLC NAND will indeed gradually migrate to TLC," the Counterpoint analyst noted. Unlike traditional DRAM, which remains crucial in long-lifecycle infrastructure like automotive, MLC NAND has been surpassed in performance by SLC and replaced in cost-efficiency by TLC and QLC, seeing its competitive advantage compressed from both sides.
"The downstream market has already begun migrating to TLC and QLC to reduce costs. As upstream MLC supply continues to tighten, this substitution trend is expected to accelerate further." However, the analyst also mentioned that the resulting increase in BOM (Bill of Materials) costs will squeeze downstream manufacturers' profits, with industrial and embedded suppliers serving network equipment manufacturers and large OEMs being the most affected.
TrendForce also mentioned that in the long term, the growth potential of MLC NAND's main end-markets is limited. Furthermore, if some applications accelerate the adoption of enhanced TLC solutions, or if the overall NAND Flash market sentiment reverses significantly, MLC product prices could still face indirect pressure.
From a "commodity" to a "niche luxury," the changing fate of MLC NAND is a microcosm of resource reallocation in the memory industry during the AI era. With limited capital expenditure and cleanroom space, upstream manufacturers are decisively shifting capacity towards higher-margin products like HBM and high-layer-count 3D NAND. Meanwhile, mature memory products that once supported the stable operation of industries like automotive, industrial control, and medical are being phased out more rapidly due to their lower unit output value.
This trend is difficult to reverse in the short term, forcing downstream end-markets to accelerate technological substitution and supply chain restructuring.
For domestic memory manufacturers, the key to gaining a firm foothold in this reshaping global landscape will depend on whether they can win customer trust with stable supply capabilities and reliable product quality in the gaps left by overseas giants' exits, and find breakthroughs in independent innovation amidst patent barriers.
Comments