Morgan Stanley: Storage chips may need to digest crowded positioning in the short term, while long-term AI demand continues to support an upward cycle
AI summary card
Morgan Stanley: Storage chips may need to digest crowded positioning in the short term, while long-term AI demand continues to support an upward cycle
The report summarizes the core contradiction in the current Asia-Pacific storage segment as whether AI compute is oversupplied, why LTA has not yet driven a valuation re-rate, and whether the cycle is peaking or extending, maintaining a long-term bullish stance while warning that stock performance may weaken before earnings.
- Storage remains a cyclical industry, with year-over-year price, inventory, and earnings revision breadth near elevated levels, so short-term momentum may weaken.
- The report does not believe the cycle has ended, viewing the near-term correction as a necessary rebalancing within a structural AI bull market.
- The debate that AI compute being sold out indicates supply overhang may be valid, but it may simply reflect hyperscalers monetizing infrastructure and improving return on AI capex.
- LTA has not driven valuation re-rating because markets fear that LTA could be renegotiated as in the past, or eventually force customers to absorb undesired inventory.
- In allocation terms, the report prefers the flow of real AI-related capital and where bottlenecks are emerging, favoring DRAM and traditional memory over NAND and memory module manufacturers.
Report interpretation
Overview
This report is Morgan Stanley’s forward-looking earnings and thematic discussion on the Asia-Pacific storage chip segment, focusing on SK hynix and Samsung Electronics, and framing three investor-priority issues as the core narrative: whether the largest AI spenders are now selling compute that can be monetized, why long-term supply contracts (LTA) have not prompted a valuation re-rate in storage, and whether the cycle is at a peak or in a phase of extension. The overall conclusion is that storage may correct in the short term due to crowded positioning, high earnings revision breadth, and hyperscaler pressure, while AI capex and Agentic AI continue to support a long-term structural bull market.
Core views
Key points include: first, storage segment year-over-year price, inventory, and breadth of earnings revisions are nearing a high point in the cycle rate of change, so stock performance could enter consolidation or a phase of reset. Second, reports of AI compute being sold may heighten concerns about AI overbuilding, but may also represent cloud operators monetizing infrastructure to lift capex returns; the true validation point is 2Q26 earnings and whether hyperscalers maintain or raise capex guidance. Third, LTA has not driven immediate rerating because the market still remembers that LTA has been renegotiated historically or led to absorption of non-essential inventory; however, if AI demand remains strong, current storage LTAs could be more structural. Fourth, the report prefers DRAM and traditional memory and is relatively underweight NAND and memory module companies.
Analysis framework
The report combines catalyst event preview, mapping of investor debates, cycle-positioning assessment, earnings revision and valuation comparison, memory-price outlook, and residual income valuation framework. It does not only track storage companies’ own commentary, but emphasizes hyperscaler earnings, capex guidance, token usage trends, open-source model substitution, monetization of AI infrastructure, and the impact of crowded storage positioning on storage stocks.
Methodology notes
Residual income valuation model
In the valuation section, the report mentions using a residual income model, assuming a cost of equity of 11.5%, risk-free rate of 5%, equity risk premium of 6.5%, beta of 1.0, and terminal growth rate of 3%.
Rate of change cycle analysis
The report uses year-over-year price, inventory, and earnings revision breadth to judge whether the storage cycle’s rate of change is near a high point, and on that basis suggests short-term stock momentum may weaken.
Catalyst event preview
The report highlights SK hynix 2Q26 earnings on July 29, 2026 and Samsung Electronics preliminary 2Q26 earnings on July 7, 2026 as highly important and expects both to align broadly with market expectations.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- SK hynix (000660.KS)Core storage beneficiary and 2Q26 earnings catalyst name
- Strengths
- Benefits from AI capex, strong DRAM demand, LTA commitments, and rising storage cycle; the report discloses its rating as O/Overweight.
- Weaknesses
- In the short term it faces crowded storage-segment positioning, high earnings revision breadth, and stock pressure before earnings.
- Comparison
- Compared with NAND and memory module manufacturers, the report prefers DRAM and bottlenecks where actual capital is flowing, and SK hynix aligns more closely with that preference.
- Risks
- Terminal demand below expectations, DDR5 competition intensifying leading to over-investment on the supply side, and elevated inventories at cloud and Chinese smartphone customers.
- Samsung Electronics (005930.KS)Representative Korea Tech and storage-cycle stock, 2Q26 preliminary earnings catalyst
- Strengths
- Benefits from the storage upcycle, AI and hyperscale data center growth, technological progress, and potential improvement in capital returns; disclosed rating is O/Overweight.
- Weaknesses
- More diversified business model with exposure to product-cycle swings, competition from Apple and Chinese smartphones, and semiconductor profit concentration-driven volatility.
- Comparison
- Compared with SK hynix, Samsung Electronics is simultaneously exposed to storage, smartphones, displays, and a wider set of businesses, so storage sensitivity may be moderated by diversification.
- Risks
- Terminal demand weakness, down cycles in products and storage, competition from Apple and Chinese smartphones, and unexpected technology cadence changes.
- DRAMStorage subsegment preferred in the report
- Strengths
- Supported by AI spending and memory bottlenecks, strongest earnings revision performance; the report prefers DRAM and traditional memory.
- Weaknesses
- Year-over-year price and earnings revision breadth near historical highs, so short-term rate of change may be peaking.
- Comparison
- More preferred than NAND in the report.
- Risks
- Cyclical correction, supply expansion, and over-investment driven by intensified DDR5 competition.
- NANDA storage subsegment with lower relative preference
- Strengths
- Some price forecasts remain revised upward, such as 3Q26E Total NAND Flash raised to up 10-15%.
- Weaknesses
- The report explicitly states it prefers DRAM and traditional memory more, so NAND is not the top-priority segment.
- Comparison
- Less preferred than DRAM and traditional memory, but stronger than the least preferred memory module manufacturers.
- Risks
- Price recovery may not be sustained, terminal demand weakness, and slow inventory digestion.
Key data
- Report date2026-07-06 07:34 PM GMTDisclosure date on cover page.
- Sector viewAttractiveS. Korea Technology sector view.
- Long-term earnings outlook2027e earnings growing over 35-40%The report uses this as one basis for a long-term bullish view on storage and the structural AI capex bull market.
- Current price of Samsung ElectronicsW318,000Reference stock price disclosed in the report.
- Current price of SK hynixW2,343,000Reference stock price disclosed in the report.
- SK hynix 2Q26 catalyst2026-07-29; importance Very High; expectation In-lineThe report expects 2Q26 operating profit to be close to consensus, around W65tn.
- Samsung Electronics 2Q26 preliminary catalyst2026-07-07; importance Very High; expectation In-lineThe report expects 2Q26 operating profit of about W85tn.
- Samsung Electronics valuation assumptionscost of equity 11.5%; terminal growth 3%; 2027e P/B around 2xThe report states the target price corresponds to about 2x 2027e P/B, consistent with commodity-cycle peak expectations of around 2.0x.
- NAND price forecast direction3Q26E total NAND Flash revised from up 8-13% to up 10-15%The table shows upward revisions in some NAND sub-segment price forecasts, such as enterprise SSD and client SSD.
Impact & implications
For investors, the report implies that the storage segment should not be labeled as ended just because of short-term volatility, but also that crowded positioning and a peak in rate of change should not be ignored as stock headwinds. If 2Q26 hyperscaler earnings and capex guidance remain stable or are raised, the near-term correction could become a better entry point for storage stocks; if capex is cut, the narrative of AI compute oversupply may persist and keep pressure on storage shares. Within the industry, attention should focus more on bottlenecks in DRAM, traditional memory, and AI capital flows rather than NAND or memory module companies.
Risks
- If hyperscalers cut capex in 2Q26 or 3Q26 guidance, the AI compute oversupply narrative may continue.
- Token minimization, low-cost open-source LLMs, and enterprise AI orchestration layers could pressure frontier model and AI infrastructure demand expectations.
- Storage positions are crowded; high net exposure is harder to maintain when volatility rises.
- DRAM earnings revision breadth is near historical highs, and there is uncertainty around the room and pace for further upward revisions.
- Terminal demand below expectations could compress ASP and earnings.
- Intensifying DDR5 competition could drive over-investment on the supply side.
- If cloud and Chinese smartphone customers continue to hold high inventories, demand and prices could be dragged down.
- Morgan Stanley has potential conflicts of interest with investment banking relationships at some covered companies; investors should assess carefully based on disclosures.
What to watch
- SK hynix 2Q26 earnings on July 29, 2026, and management commentary on 3Q26 storage prices, LTA, and capex.
- Samsung Electronics preliminary 2Q26 results on July 7, 2026, especially whether approximately W85tn operating profit expectation is met.
- Whether hyperscalers maintain or raise AI capex in 2Q26 results.
- 2Q26 earnings and 2H26 guidance of token providers; monitor the impact of a shift from token maxing toward token minimization.
- Open-source LLMs, especially low-cost Chinese models, and their substitution effect on frontier model usage and AI infrastructure demand.
- Spot and contract price trends for DRAM and NAND, especially whether year-over-year price change continues to ease.
- Relative rotation of storage shares versus broader indices and other semiconductor subsegments such as MLCC and semicap.
- Whether LTA execution reflects structural demand or repeats historical patterns of renegotiation and passive inventory absorption.