On July 21, 2026, TrendForce published a forecast stating that the NAND Flash supply shortage will move toward easing in the second half of 2027. According to the forecast, supply will fall short of demand by 4-5% in 2026, but by the second half of 2027, the growth in supply bits will outpace demand growth. This shift will not be driven by a wave of new large-scale fabs coming online all at once during 2027. Rather, it will result from existing lines converting to higher-density generations, Chinese manufacturers expanding equipment, and production of smartphones and notebook PCs shrinking in the face of high prices. Accordingly, the second half of 2027 should not be understood as a "deadline for resolving the shortage," but rather as the point when the overall NAND market begins to regain balance.
The Conditions Under Which a 4-5% Shortage Turns Positive
According to TrendForce's estimates, in the 2026 NAND market, supply bits will fall 4-5% short of demand. NAND manufacturers' inventories will remain low, and only module makers—facing a slump in the consumer market—will be building up inventory. Other buyers' inventories will remain generally at controlled levels, meaning this is not a phase where excess supply accumulates and supply-demand naturally loosens. Even so, TrendForce sees supply exceeding demand by the second half of 2027.
What this forecast measures is the total number of NAND bits produced and consumed globally. Enterprise SSDs and PC SSDs differ both in which generations they adopt and in supply prioritization. The conditions are likewise not the same for smartphone-oriented Universal Flash Storage (UFS) and embedded applications. Even if the aggregate supply becomes a surplus, this does not necessarily mean all products will simultaneously become easier to obtain.
The same time lag applies to pricing. TrendForce had forecast that NAND contract prices in Q2 2026 would rise 70-75% quarter-on-quarter. In an environment where manufacturer inventories are low and cloud providers are locking in long-term supply contracts, it cannot be said that contract prices or retail SSD prices will drop the moment supply exceeds demand. The date of late 2027 is not a deadline for price normalization, but rather a benchmark for when the forces that have been driving the supply shortage begin to weaken.
Increasing Capacity Per Wafer Ahead of New Fabs
The primary means of increasing supply from 2026 through 2027 will not be a major expansion of wafer input volume, but rather migration of the manufacturing process. 3D NAND stacks memory cells vertically while also increasing planar density, allowing more bits to be extracted from the same wafer. TrendForce predicts that because space at existing fabs is limited and manufacturers are prioritizing DRAM production increases, companies in South Korea, the United States, and Japan will continue to update existing lines and selectively expand certain fabs.
On July 3, Kioxia and Sandisk began production of their 10th-generation 3D flash memory at the second manufacturing building of the Kitakami Plant in Iwate Prefecture. The 1Tb TLC product, for which sample shipments were announced the same day, has 332 layers, representing a 59% increase in bit density compared to the 8th generation's 218 layers. However, these specifications are for functional verification samples and may differ in mass-produced products. The company plans to increase capital expenditure for fiscal year 2026 to 450 billion yen, up from 280 billion yen the previous fiscal year, and to invest a similar amount in fiscal year 2027. Through capacity expansion of the 8th generation and the launch of the 10th generation, the company states this will support an average annual growth rate of 22% on a capacity basis.
In South Korea, SK hynix began mass production in August 2025 of Quad-Level Cell (QLC) NAND with 321 layers, storing 2Tb (terabits) per chip. Because QLC records 4 bits per cell, it increases the capacity obtainable from the same area. As this kind of layer-stacking and multi-level-cell progress continues, supply bits will grow without waiting for new fab completions.
Meanwhile, the new NAND wafer fab that Micron has begun constructing in Singapore is a large-scale project involving approximately $24 billion in investment over ten years, but wafer production is scheduled to begin in the second half of 2028. At minimum, this fab is not equipment that will directly support the restoration of balance in the second half of 2027. The short-term improvement that TrendForce envisions depends on plans to extract more capacity from existing assets.
Why Supply-Demand Loosens Even as AI Server Demand Grows
Servers already account for over 40% of global NAND bit demand. Next-generation server platforms from Intel and AMD will launch in the second half of 2026, and TrendForce forecasts that server shipments will grow 17% year-on-year that year. In 2027, the practical deployment of agentic AI is expected to support general-purpose server investment, with easing CPU supply constraints and long-term memory contracts also boosting shipment growth. The premise is not that AI-related storage demand will weaken.
Even so, the overall market is moving toward balance because smartphones and notebook PCs together still account for approximately 40% of NAND bit demand. TrendForce projects that smartphone production in 2026 will decline 15-20% year-on-year, and notebook PC shipments will fall approximately 10%. For 2027, the forecast anticipates continued contraction in both categories—smartphones as consumers become more price-sensitive, and notebook PCs as component costs such as memory and CPUs remain elevated. Even if premium-priced AI smartphones support part of the market, they are not expected to be powerful enough to push overall unit volumes back into growth.
While supply bits steadily increase, the number of bits consumed by consumer devices is unlikely to grow much. Even as server demand expands, the contraction in the remaining major markets holds down total demand, allowing supply growth to catch up by the second half of 2027. This supply-demand improvement is a forecast premised on both increased production and decreased demand occurring simultaneously.
China's 19% Share, and the Distance Before Prices Move
Another factor affecting the supply side is Chinese manufacturers. TrendForce forecasts that as new manufacturing equipment comes online, the share of global NAND bit production held by Chinese manufacturers will rise to approximately 19% by 2027. While the published materials do not provide a company-by-company breakdown, there is a difference in approach: manufacturers in South Korea, the United States, and Japan are selectively expanding existing equipment, whereas Chinese manufacturers are growing supply volume through the deployment of new equipment.
However, a change in supply share does not directly translate into lower prices for consumer SSDs. NAND manufacturers prioritize enterprise SSDs, and cloud providers secure their required volumes through long-term contracts. Depending on which products the new supply bits are allocated to, the timing of improvement will differ between enterprise SSDs and PC SSDs.
What needs to be confirmed in the second half of 2027 is not only whether total bit supply exceeds demand, but also whether low manufacturer inventories recover and whether supply allocation to PCs and smartphones increases. Only once these conditions align together will the easing of the NAND shortage become visible in the pricing and availability of consumer SSDs.