AI Demand and Supply Constraints Are Driving NAND into a New Cycle of High Prices and High Profits

Overview

The report argues that the NAND market is shifting from a traditionally highly cyclical industry to a new phase jointly driven by AI demand, long-term supply agreements, and persistent supply constraints. Agentic AI increases the number of server CPUs, inference tasks expand enterprise SSD deployments, and rack-level pooling and disaggregated architectures further increase flash density per accelerator. At the same time, capital expenditure cuts over the past several years, prioritization of DRAM/HBM allocation, and the greater cleanroom space requirements of advanced node migrations make it difficult for wafer capacity to grow rapidly. As a result, the long-term midpoint for prices, revenue, profit per wafer, and free cash flow is expected to be significantly higher than historical levels.

Core views

Core conclusions include: AI data centers are becoming a structural incremental source of NAND demand; long-term supply agreements improve price and demand visibility and reduce cyclical volatility; suppliers with higher enterprise SSD exposure can obtain average selling price premiums; supply discipline and physical capacity constraints will keep industry shortages lasting longer; the industry's long-term operating margin is expected to approach 60%; the Americas, driven by hyperscale data center spending, are expected to surpass China in 2026 to become the largest NAND market; agentic AI, nearline SSDs, physical AI, HBF, and rack-level memory expansion may still bring upside demand beyond forecasts.

Analysis framework

The report combines quarterly supplier operating data, prices and shipments, wafer output and utilization, capital expenditures and wafer fabrication equipment investment, enterprise SSD prices across technologies, revenue and profit per wafer, regional revenue structure, and supplier market share for supply-demand modeling, and compares the latest forecast with the previous forecast quarter by quarter to form the industry revenue and profitability outlook for 2026 to 2031.

Methodology notes

• Industry supply-demand analysisNAND Supply-Demand Balance Framework

  Dynamic matching of demand growth and effective wafer supply

  Assesses the degree of market shortage and price direction by combining demand from AI data centers, enterprise SSDs, and consumer electronics with wafer output, capacity utilization, node migration, and capital expenditure.

• Cycle analysisMemory Cycle Framework

  Cyclical feedback among prices, production, shipments, and margins

  Compares NAND price changes, production adjustments, bit shipments, and supplier profits to evaluate differences between this cycle and historical cycles, as well as whether the profit midpoint has structurally shifted upward.

• Scenario forecastingAI Demand Scenario Analysis

  Estimating incremental NAND demand based on the evolution of AI architectures

  Incorporates applications such as KV cache spillover, vector retrieval, persistent memory, rack-level flash, nearline SSDs, and HBF into demand assumptions, while highlighting that rapidly changing AI architectures may cause forecast volatility and historical data restatements.

• Competitive landscape analysisSupplier Share and Unit Economics Comparison

  Comparing revenue share, bit shipment share, and profit per wafer

  Covers Samsung, Kioxia, SanDisk, Micron, SK hynix, Solidigm, and YMTC, analyzing the impact of product mix, enterprise SSD exposure, and regional market differences on revenue and profit share.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Samsung, Kioxia, SanDisk, Micron, SK hynix, and Solidigm

  Major global NAND suppliers, directly affected by changes in industry prices, shipments, and margins.

  Strengths

  Improved supply discipline, growth in AI data center demand, an increase in long-term supply agreements, and an upward shift in the price midpoint are positive for revenue, profit per wafer, and cash flow.

  Weaknesses

  Node migration uses more cleanroom space, limiting the pace of capacity expansion; enterprise SSD product mix and execution capabilities differ across vendors.

  Comparison

  Suppliers with higher enterprise SSD exposure typically enjoy average selling price premiums, while Solidigm's data-center-focused strategy brings higher revenue per wafer.

  Risks

  Aggressive capacity expansion, weaker-than-expected AI demand, customer inventory adjustments, or deteriorating price elasticity could all weaken profitability improvement.

• YMTC

  A major Chinese NAND supplier and an important source of global supply growth and competitive landscape change.

  Strengths

  Its revenue share exceeded 10% in 2025, and it has gained significant share in the Chinese market.

  Weaknesses

  Some industry operating and capacity statistics do not include YMTC, making comparable data less complete than for other major suppliers.

  Comparison

  YMTC's share is rising, while Samsung remains the global leader but its lead has narrowed.

  Risks

  Continued capacity expansion could ease global supply tightness and put pressure on medium- to long-term prices and other suppliers' shares.

• Enterprise SSDs and AI Data Center Storage

  The core application area for this round of NAND demand re-rating.

  Strengths

  KV cache spillover, vector retrieval, persistent memory, and rack-level flash significantly increase SSD configurations per server and per accelerator.

  Weaknesses

  Demand assumptions are highly dependent on constantly changing accelerator architectures, rack designs, and deployment pace.

  Comparison

  Compared with traditional server-attached storage, rack-level pooling and disaggregated architectures extend NAND demand into higher-density and broader storage tiers.

  Risks

  Architectural changes, improvements in software efficiency, or deployment delays could lead to downward revisions in demand forecasts and historical data restatements.

• NAND Wafer Fabrication Equipment and Advanced Materials Suppliers

  Beneficiaries of a recovery in NAND capital expenditure and materials-driven expansion of 3D NAND.

  Strengths

  Infrastructure expansion, advanced node migration, and more complex 3D NAND processes support demand for equipment and materials.

  Weaknesses

  Capital intensity in this cycle remains significantly lower than in historical cycles, and short-term wafer capacity expansion is constrained by cleanroom space.

  Comparison

  Lam Research has maintained the largest share of the NAND wafer fabrication equipment market in recent years, and U.S. equipment vendors as a whole hold a dominant position.

  Risks

  Suppliers maintaining capital discipline, prioritizing investment in DRAM/HBM, or delaying expansion could limit equipment order growth.

• Smartphone Supply Chain

  The downstream cost bearer of rising NAND prices.

  Strengths

  Long-term increases in storage capacity per device continue to provide structural bit demand.

  Weaknesses

  NAND bill-of-materials costs may be significantly higher than the long-term range of approximately 4% to 6% of total device bill-of-materials cost.

  Comparison

  Compared with AI data centers, smartphone demand is more price-sensitive and has weaker cost pass-through capability.

  Risks

  Rapid increases in storage prices may compress terminal manufacturers' margins, slow the pace of configuration upgrades, and suppress shipment demand.

Key data

• NAND market size in the first quarter of 2026$45.3 billionUp 89% quarter over quarter, reaching an all-time high.
• Change in average selling price in the first quarter of 2026Up more than 90% quarter over quarterThe main driver of the substantial increase in market revenue for the quarter.
• Industry operating margin in the first quarter of 2026Approximately 63%Data center demand and tight supply pushed industry financial performance to a record level.
• Long-term operating margin frameworkNearly 60%The long-term framework used in the previous report was approximately 40%.
• YMTC revenue shareExceeded 10% in 2025Samsung remains the market leader, but its lead has narrowed.
• Recovered equipment capacityApproximately 75% of the pre-production-cut levelAlthough utilization has recovered, node migration and cleanroom space constraints are pressuring overall wafer capacity.

Impact & implications

The industry profit pool may shift from a short-term price rebound to a more durable structural expansion. Suppliers with higher enterprise SSD exposure, the ability to participate in AI data center growth, and product pricing power are more likely to obtain premiums in revenue and profit share; NAND wafer fabrication equipment and advanced materials vendors may also benefit from a subsequent recovery in capital expenditure. Downstream price-sensitive markets such as smartphones face risks from rising bill-of-materials costs and suppressed demand. Since the industry is already operating near full capacity, incremental demand is more likely to translate into prices rather than shipment growth.

Risks

• AI infrastructure is evolving extremely rapidly, and changes in accelerator architecture, rack design, and deployment assumptions may lead to substantial forecast volatility and historical data restatements.
• Rapid increases in NAND prices may cause smartphone storage bill-of-materials costs to significantly exceed the long-term range, thereby suppressing end-market demand.
• Continued expansion by YMTC or faster capital expenditure by major suppliers may cause supply growth to exceed current forecasts and depress prices.
• Prioritization of DRAM/HBM allocation, insufficient cleanroom space, and advanced node migration may limit NAND output while increasing supplier execution risk.
• Suppliers' accounting treatments for idle capacity charges, inventory write-downs, and reversals are inconsistent, creating definitional differences in cross-sectional comparisons of industry margins.
• Some statistics cover only Samsung, Kioxia, SanDisk, Micron, SK hynix, and Solidigm, excluding YMTC and other vendors, which may underestimate total industry size.
• Free cash flow forecasts have not yet been disclosed, and quantitative conclusions on cash generation capacity lack full disclosure.

What to watch

• Quarterly changes in NAND average selling prices and prices for enterprise TLC, QLC SSDs, and raw components.
• The number of AI server CPUs, SSD density per accelerator, and progress in rack-level flash deployment.
• Coverage, duration, and pricing mechanisms of long-term supply agreements.
• Wafer output, capacity utilization, inventory levels, and restarts of idled capacity at major suppliers.
• NAND capital expenditure, cleanroom expansion, advanced node migration, and wafer fabrication equipment investment.
• Changes in revenue and bit shipment shares of Samsung, Kioxia, SanDisk, Micron, SK hynix, Solidigm, and YMTC.
• Whether hyperscale data center spending in the Americas can make it the largest NAND regional market in 2026.
• The impact of a rising share of NAND bill-of-materials cost in smartphones on device configurations and demand.