Memory Prices Increase ~60% QoQ in 2Q26; Growth to Slow in H2

Overview

This Bernstein May global memory price tracking monthly report concludes that DRAM and NAND contract prices rose sharply (~60% QoQ) in 2Q26, surpassing prior model expectations. The surge was primarily driven by robust AI-related demand in the Server segment and supply shortages in the Mobile segment. However, the report cautions that weakening consumer demand will significantly slow the pace of price increases in H2, with prices likely peaking and beginning to decline in H2 2027.

Core views

2Q26 pricing surge exceeded expectations, led by Server and Mobile: In May, contract prices rose slightly QoQ, resulting in a cumulative 2Q26 DRAM contract price increase of +64% vs. 1Q26 (+46% for PC, +53% for Server, ~+80% for Mobile, ~+85% for Consumer), while the NAND blended contract price rose ~+60% (primarily driven by ~+70-80% increases in Mobile NAND and SSD; wafer prices rose only ~+25%).Spot prices stabilized and began rebounding in May: PC DDR4/5 spot prices rose +5–7.4% QoQ, Server spot prices rose +1.6–3.1%, and NAND wafer spot prices rose +0.2% QoQ — indicating prior sell-offs have been absorbed. Strong Server demand vs. Demand Destruction in Mobile: Server DRAM demand remains robust, absorbing capacity from delayed HBM4 releases and reallocated supply. DDR5 now accounts for >90% of server DRAM, and U.S. cloud service providers (CSPs), enjoying priority allocation, have largely completed their LTA negotiations. In contrast, after earlier price surges, smartphone OEMs are cutting production runs and memory configurations, expecting a 16% YoY decline in smartphone shipments this year. As a result, Mobile DRAM price increases in H2 are expected to slow to 0–10% QoQ. Although consumer-tier demand remains relatively resilient for now, it is expected to weaken, and some OEMs have already begun down-specing to control costs. Pricing slowdown in H2; peak expected in H2 2027: The report maintains that the current shortage will persist into 2027, but expects the pace of price increases to decelerate notably in 3Q26, normalizing and turning downward in H2 2027 as new capacities come online. Broadening LTA coverage may help smooth the transition. NAND is viewed with greater caution, primarily due to intensified competition from China.

Analysis framework

The analysis follows two main strands: (1) the spot-to-contract price transmission mechanism, and (2) demand disaggregation by end-market. First, spot price stabilization and rebound serve as leading indicators for further contract price increases. Second, memory demand is segmented by PC, Server, Mobile, and Consumer, tracking segment-specific contract price changes, OEM behavior, and capacity allocation. By comparing TrendForce’s May contract and spot data with the firm’s internal model, the report affirms the 2Q26 pricing overshoot and结合LTAs negotiation progress and capacity ramp plans, projects a slowdown in price growth and a bottom/turnpoint in 2H27.

Methodology notes

• Industry/sector analysis frameworkSupply-demand framework

  Supply-Demand Framework

  The report analyzes增量 supply sources and absorption capacity across end-markets to assess shortage sustainability and pricing momentum.

• Industry/sector analysis frameworkPrice-Volume Decomposition

  Price-Volume Decomposition

  DRAM and NAND prices are decomposed by application, with weighted average price changes used to compute sector-wide ASP shifts — avoiding distortion from single-product volatility.

• Industry/sector analysis framework

  Spot-to-Contract Price Transmission Mechanism

  Spot prices are treated as leading indicators of contract price moves; stability and recovery in spot prices suggest prior sell-offs are resolved and contract price momentum can resume.

• Cyclicality & Sentiment FrameworkCycle Turnpoint Analysis

  Cycle Turnpoint Analysis

  By tracking OEM production cuts, down-specing, and LTA negotiation progress, the report identifies the transition from 'severe supply shortage → rapid price hikes' to 'weakening demand → slowing price increases', forecasting a price peak and reversal in H2 2027.

Asset mapping & comparison

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

• Samsung Electronics

  Memory leader; benefits from early Server DRAM supply access and Mobile DRAM price increases

  Strengths

  Strong pricing influence and flexible capacity reallocation in 2Q26

  Risks

  Early end of favorable pricing environment; progress of Chinese NAND makers

• SK hynix

  Key beneficiary of HBM and Server DRAM demand

  Strengths

  Leading HBM technology; primary DDR5 supplier to Server market

  Weaknesses

  Initial Mobile DRAM pricing requests were relatively modest (55–60%)

  Risks

  Early end of favorable pricing environment; progress of Chinese NAND makers

• Micron (MU)

  Beneficiary of Server and Mobile DRAM pricing increases

  Risks

  Early end of favorable pricing environment; progress of Chinese NAND makers

• KIOXIA

  Primarily wafer-focused; seen as the weakest performer in price rallies and under competitive pressure from China

  Weaknesses

  Structurally weaker NAND positioning; facing YMTC competition

  Comparison

  weakest pricing gains vs. peers (wafer only up ~25%)

  Risks

  NAND demand misses expectations; weaker-than-expected policy support in Japan

• SanDisk (SNDK)

  Highly benefited from strong SSD pricing

  Strengths

  SSD pricing ~+70% locked in high ASP

  Weaknesses

  Recent high data面临 cyclical downturn; investor communication appeared disorganized

  Risks

  NAND weakness may become structural rather than cyclical

Key data

• 2Q26 DRAM contract price QoQ change64%Overall; +46% for PC, +53% for Server, ~+80% for Mobile, ~+85% for Consumer
• 2Q26 NAND blended contract price QoQ change~60%Mobile NAND & SSD: +70–80%; Wafer: +~25%
• 5-month PC DRAM spot price QoQ change5–7.4% (DDR4/5)Spot price rebound reflects recovery after prior sell-offs
• 5-month Server DRAM spot price QoQ change1.6–3.1%Large gap between spot and contract prices indicates severe supply shortage
• DDR5 share in Server DRAM (2Q26)>90%DDR5 has become server mainstream; DDR4 faces EOL shortage, though its share is marginal
• Full-year smartphone shipment growth forecast-16%OEMs cutting production and memory configurations; Mobile pricing growth expected to slow to 0–10%/Q in H2

Impact & implications

For memory manufacturers, the stronger-than-expected 2Q26 pricing will lock in exceptionally high ASPs, fueling strong near-term earnings. However, decelerating price increases in H2 imply weakening QoQ earnings momentum, with LTAs potentially smoothing the transition. Structurally,DRAM looks more favorable due to AI tailwinds, while NAND faces long-term headwinds from Chinese competition. The expectation of a peak in the pricing cycle may already compress valuation multiples ahead of time.

Risks

• Early termination of favorable pricing environment (weakening demand or rising supply)
• Investor sentiment and valuation compression risk
• Progress of China in memory — especially NAND — poses downside risk
• Consumers’ demand destruction may be deeper than currently anticipated
• NAND weakness could become structural rather than cyclical (specific to SanDisk)

What to watch

• Whether 3Q26 contract price increases slow as expected (DRAM ~10–20% QoQ)
• Outcome and scope of LTAs negotiations with Chinese CSPs
• Actual execution of OEM down-specing and production cuts in consumer segment
• Impact of delayed HBM4 ramp on Server DRAM supply dynamics
• Timeline of new capacity ramp and validation of the 2H27 price peak hypothesis