AI Memory & HBM
The performance bottleneck for AI accelerators has shifted from compute to memory bandwidth. HBM (High Bandwidth Memory) and DDR5 demand is surging explosively, ushering in a memory super-cycle.
Investment Confidence
Theme Analysis
Key Drivers
- DDR4/DDR5 prices surged 4x, NAND QoQ +55-60% — shortage forecast through 2030 (SK Group)
- AI data centers consuming 70% of total memory production (Deloitte) — zero-sum capacity allocation
- HBM demand growing 100%+ annually, HBM TAM $100B (2028E)
- Samsung NAND wafer target cut 490M→468M, SK Hynix 190M→170M
- Chipflation — BlackRock warns CPI computer prices up 15.5% over 3 months
Risk Factors
- Samsung May strike could cut Pyeongtaek production 50% — Worst Case scenario trigger
- Qatar helium supply 30% offline — Korea imports 64.7% of helium from Qatar
- Energy cost surge → hyperscaler Capex cuts could end memory super-cycle early
- Chinese memory firms (YMTC, CXMT) emerging as alternative supply
- Smartphone/PC OEM demand destruction — IDC bear case shipments -8.9%
Catalysts
- Samsung strike early resolution → supply relief + price stabilization rally
- TSMC (4/17) earnings — AI demand guidance confirmation
- HBM4 mass production ramp — SK Hynix 70% market share forecast (UBS)
- Memory Big 3 2026 earnings surprise — pricing power strengthening
- Hormuz Strait reopening → helium supply normalization
Related Investment Ideas
4Jeju Semiconductor — The Only Listed Korean Fabless Beneficiary of Memory Big 3's Legacy Exit
As Samsung/SK hynix/Micron accelerate LPDDR4/4X EOL, Jeju Semiconductor sits as the only listed Korean fabless play on the resulting low-density LPDDR supply void. Capital-light fabless model plus a mix shift toward high-margin automotive/IoT revenue drives structural re-rating. 2025Q3 revenue +197% YoY, operating profit +762% YoY confirms the turnaround.
SK hynix — Global #1 in HBM, Top Beneficiary of the Memory Super Cycle
Absolute #1 with 50%+ global market share in HBM3E. HBM4 mass production ready. Operating profit forecast of KRW 100T+ for 2026. Dominant supply position for NVIDIA-bound HBM continues.
Micron — HBM3E Shipments Ramping, America's Memory Champion
Full-scale HBM3E production driving rapid increase in high-margin product mix. Data center revenue share exceeds 50%. CHIPS Act subsidies funding U.S. manufacturing expansion.
SanDisk — The Purest Beta on NAND Re-rating in the AI Era
A pure-play on NAND flash. The AI inference storage cycle drove FY26 Q3 revenue +251% YoY and a 78.4% non-GAAP gross margin. The Kioxia JV is extended to 2034, and HBF (High-Bandwidth Flash) adds option value.
Related Reports
62Weekly Market Review: Week 28 (July Week 2)
The Week Samsung Posted Record Earnings and Still Crashed -7% — a Peak-Out-Fear Circuit Breaker, a Two-Day -10% Plunge and a Two-Day Rebound, and SK Hynix's $26.5 Billion Nasdaq IPO, the Largest Ever by a Foreign Company
Jeju Semiconductor — When the Province's Export Data Is the Company's Earnings Sheet
Legacy-memory fabless Jeju Semiconductor (080220): under shipper-location accounting, Jeju's chip exports are effectively this single company's gauge. We read earnings through that stand-in first, then confront the gap between the export read and the share price
Who Pays for Memory — Four Payment Channels and a Redrawn Contract Map
The buyer map and the contract map — dissecting the AI memory bill down to channel, company, and contract. The thinnest point on this map is that the largest new buyer, which reserved 40% of the world's DRAM, is a loss-making company
Documents, Not Shovels — The Stock-Price Grammar of Memory Capacity Cycles
How to read the Korea–US–Japan mega-expansion wave and the early-July crash — what broke the stocks was not groundbreaking but documents combined with a demand crack, and a contract revolution is rewriting that grammar
CXMT Is a Gauge, Not a Threat — What China's DRAM IPO Measures About the Memory Shortage
+719% revenue is a shortage reading, not an invasion — the Tencent deal disarmed Apple's China card; a bullish confirmation bounded at FY27, with FY28 on a separate watch track
Weekly Market Review: Week 27 (July Week 1)
The Week KOSPI Crashed -7.89% in a Day and Rebounded +5.76% the Next — the Meta Cloud Shock, a 17-Year High in USDKRW at 1,555.8, a Payrolls Shock, and a Buy Sidecar
Weekly Market Review: Week 26 (June Week 4)
From an all-time high of 9,114 to two circuit breakers — KOSPI swung nearly 7% in a single week, Micron's earnings beat confirmed the memory supercycle, and the AI capital cycle kept flowing regardless of the short-term collapse
Higher HBM Mix, Safer Stock? — How Base Die Logic Migration Shifts AI Memory Margin to Foundries
From HBM4, the base die becomes a leading-edge logic chip. Part of AI memory's value migrates from the three memory makers to foundries. The same 'HBM mix' hides different profit structures at Samsung, SK Hynix, and Micron
SK hynix Nasdaq ADR — Anatomy of a 43 Trillion Won Deal and the Premium Equation
The memory champion debuts on Nasdaq as an American depositary receipt on July 10. The largest US listing by a foreign company in history is aiming not at fundraising but at a re-rating — the valuation gap with Micron, and the 2.5% conversion cap separating the underlying shares from the ADR, together write the premium equation.
SK Square — Buying HBM Leader SK Hynix at a Discount
A holding-company lever on SK Hynix, the true global leader of the memory supercycle. About 98% of net asset value (NAV) is a single stock, SK Hynix, and a narrowing holdco discount makes that exposure even cheaper.
The Pulled-Forward Future — The AI Capital Cycle's Repayment Chain and the Accounting of Inference Margins
Demand (PC, memory) and capital (bonds, equity) both dragged into the present — the whole chain's fate rests on the durability of the inference margin
Tokyo Electron — The 90% Coater Monopoly Every EUV Wafer Must Pass
About 90% global share in coater/developer track tools — the gating step for ASML's EUV patterning and Japan's #1 equipment maker
KLA — The Monopoly Tollgate of Inspection, Where Shrinking Equals Revenue
58% process-control share; node and HBM complexity raise inspection counts per chip — the gate on yield
Arista Networks — The #1 Back-End Ethernet That Stitches AI Clusters Together
In the AI back-end fabric that ties tens of thousands of accelerators, Ethernet is pushing out InfiniBand — and Arista sells that system
Applied Materials — The Full-Line Pickaxe That Now Sells Packaging Too
The deposition-etch-CMP leader opens a second growth track in HBM-stacking advanced packaging
The EDA Duopoly — The Design Gate Every AI Chip Passes Through: Cadence and Synopsys
TPU, Trainium, MI400 — no custom AI silicon reaches tape-out without these two companies' design software
Weekly Market Review: Week 25 (June Week 3)
The Week KOSPI Crossed 9,000 for the First Time — an HBM4E-Led Memory Rally, Warsh's Hawkish FOMC, an Intel-Apple Deal, and a Korea-Only Record High on a Juneteenth Holiday
Micron — America's One and Only HBM Card
The only memory maker headquartered on US soil that builds the HBM (high-bandwidth memory) inside AI accelerators. The 2026 memory cycle has redrawn Micron from a cyclical name into an AI-infrastructure name.
Broadcom — The Tollgate on AI Infrastructure's Road That Is Not NVIDIA's
XPU (Custom AI Accelerator) Co-Design and 102.4 Tbps Ethernet — Ruler of the Custom-Silicon Track
Seagate — Bellwether of Mass Storage AI Revived
The HAMR areal-density leader that revived the HDD industry — anatomy of TAM, cost, supply discipline, the technology war, and valuation
The CPU in the Inference Era — What Is Measured Fact and What Is Forecast
Decomposing the claim that 'inference needs this much CPU' into measured fact, stakeholder forecast and offsetting factors — three tiers of evidence and the Korean memory derivative
Weekly Market Review: Week 24 (June Week 2)
A Volatile Week Shaped by Iran Missiles and a 3-Year-High CPI — From a KOSPI Circuit Breaker (-8.29%) to an +8% Rebound, with Ceasefire Hopes Turning the Weekend
Lam Research — When Etch and Deposition Intensity Is the Revenue
Inside the structural beneficiary of the $140B wafer fab equipment (WFE) cycle, where chip verticalization raises the number of etch and deposition steps per wafer non-linearly. It is the highest beta to the AI memory supercycle transitioning into equipment orders
ARM — The Moment the IP Company Holding AI Compute's Instruction Standard Becomes a Silicon Company
The ARM architecture has captured half of AI data center CPUs. On top of the proven engine of rising v9/CSS royalty-per-chip mix, a second growth engine has been added 35 years on — its own in-house silicon AGI CPU. We dissect the compute architecture layer.
The Accounting of GW - Paper Capacity vs Physical Capacity in AI Data Centers
A site-by-site reconciliation of announced pipelines against energized capacity - delays do not destroy demand, they shift it to the right
MLCC and FC-BGA — The Benefit Structure of Two Battlefields, Finished Goods and Materials
Dissecting the AI-server component supercycle as a vertical value chain — where finished goods (Samsung Electro-Mechanics et al.) and materials each benefit
Weekly Market Review: Week 23 (June Week 1)
From Record Highs to an AI/Chip Plunge — A Week When Broadcom's Guidance and a Hot Jobs Report Hit Multiples at Once
SanDisk — The Purest Beta of NAND Memory Re-Rated in the AI Era
Pure-play NAND flash maker SanDisk (SNDK): the AI inference storage cycle, the Kioxia JV, and the HBF option in one stock
Intekplus — The 'Inspection' Node of the Semiconductor Capex Cycle
Inspection-equipment maker Intekplus (064290): a narrow, deep leader exposed to the quad expansion of foundry, packaging, and memory
Weekly Market Review: Week 22 (May Week 5)
DELL After-Hours +15% Big Beat + KOSPI 8,476 Record High — A Week When AI Infrastructure Revenue Proved Pricing Power Twice
Samsung Electro-Mechanics — The Top Korean Component Pick the Data-Center BOM Points To
Three axes — MLCC, FC-BGA, and glass substrate: why the data-center Bill of Materials (BOM) converges on a single company
Weekly Market Review: May Week 4
NVIDIA Full Beat + Samsung Strike Settlement — S&P 8-Week Win Streak, Dow All-Time High as KOSPI Surges 6.55% in Single Day
The Place Memory Took — Vera Rubin Rewrites the Data Center BOM
Within a VR200 NVL72 rack at $7.8M, memory takes 25.6% — the "HBM 15.5%" proposition evolves into "memory 25.6%"
PSK Holdings — The Small Window of Descum Is Opening
Korea's only equipment maker breaking Japan ULVAC's 30-year HBM/CoWoS Descum monopoly — a multi-exposure thesis
NVIDIA Earnings D-1: Six Bullish Theses for the Next 24 Hours
Q1 FY27 Pre-Earnings — Korea's 4-Line Supply Chain Beneficiary Check
AI Data Center BOM Reach Rate
AI Data Center Deep Dive — Part 3. NVIDIA gets a share, HBM, optical, and power must follow at theirs. Reverse-engineering revenue ceilings by industry through BOM accounting identity.
Reverse-engineering Agentic AI Penetration
Sequel to AI Data Center's Split — measuring the real depth of the agent economy through three gauges: tokens, traffic, ARR
Weekly Market Review: Week 20 (May Week 3)
KOSPI hits 8,000 and crashes -6.12% the next day — AI super-cycle fundamentals and multiple limits collapsed into a 4-hour window
When GPUs Become an Asset: AI Infrastructure's Capital Structure Gets Rewritten
Six structural shifts in asset markets, credit, and corporate valuation triggered by CME's compute futures
Surely, the throne of this new era goes to them — doesn't it?
A short note on S&P 500 IT earnings strength and the moment of revaluation for Nvidia's 2026 AI portfolio
Is Samsung the Engine Behind a KOSPI 12,000?
Samsung fundamentals re-rated → fair market cap → reverse-engineered to KOSPI
The Splitting of the AI Data Center
How Training and Inference Diverge — Three Forks of Optical Backbone, Power, and Diversification, with Category Leaders Mapped
Weekly Market Review: May Week 2
KOSPI 7,000 Breakout + Samsung's $1T Club + AMD/ARM Beats — Korea Memory's Pricing Power Exports to US
The Memory Pickaxe of This AI Gold Rush, Revisited
Why memory cycle peak debate is premature
RAMpocalypse — What Happens After the Memory Big-3 Leave
Memory Big-3 Series, Part 3 — AI is devouring the raw materials of memory; an industry golden age and consumer regression unfold side by side, in 8 questions
AI's Real Bottleneck — OSAT Packaging Wars
CoWoS capacity grew 13x and still fell short — the structural beneficiaries of the packaging bottleneck
Weekly Market Review: Week 18 (May Week 1)
Big Tech 4 earnings and META -6.15% — for now, the market only watches one CapEx number
AI Infrastructure 4 Bottlenecks — Big Tech Q1 2026 Earnings Summary
Memory · Power · Custom Chips · Optics — Four Bottlenecks Converging Simultaneously
Global Semiconductor CapEx Cycle Part 1 — Macro Decomposition
TSMC 2nm + simultaneous memory big-three expansion + the suppliers' triple order cycle
KOSPI 10,000 Path Analysis
The Triple Shortage's Korean Supply-Chain Transposition Opens the First-Ever 300T KRW Operating Profit Era
The All-Front Semi Shortage Confirmed by Intel Q1
CPU Renaissance · HBM/CoWoS/ABF Sold Out · ASIC Surge · Korean Value-Chain Exposure Map
Every Market the Memory Big-Three Are Leaving
Market Sizing · Follower Revenue/Profit · A Quantitative Re-rating Read
Weekly Market Review: Week 17 (April Week 4)
SK Hynix 72% OpMargin · Intel +2,800% Surprise — Korea-US Semis Hit Record Highs Together
The Triple Power Chain of Agentic Transition
Gas Turbines, Optical Networking, and How Efficiency Amplifies Absolute Demand
Big 3 Memory Legacy Exit & Successor Deep Dive
Who Fills the Void — Screening Korean Value Chain Second-Tier Beneficiaries
Jeju Semiconductor Just Printed ₩67B Operating Profit — What +1,713% Points To
A Korean fabless filling the LPDDR vacuum the memory three vacated. The supercycle's second-order beneficiary has arrived.
China LLM API Surge: The Token War Reversal
DeepSeek·Qwen 65% Share | 10T Daily Tokens | Price Disruption Strategy
GPU Rental Price Surge: The Great Compute Shortage
Blackwell +48% | H100 +38% | Memory Supercycle | Neocloud Boom
Samsung Just Printed ₩57T Operating Profit — Now Look at What Comes Next
The memory supercycle peak is now confirmed in numbers. Consensus average is ₩339K. The market is looking higher.
Semiconductor Shortage Deep Dive: The 'Perfect Storm' Created by AI Demand Explosion
AI Infrastructure $650B Capex, Helium Supply 30% Disrupted, Samsung Strike Risk — Structural Shortage Forecast Through 2030
Apple's Mobile DRAM Monopoly Strategy — The Hidden Winner of the Memory Supercycle
HBM Demand Explosion → Mobile DRAM Supply Shortage → Apple's Market Share Expansion Strategy Deep Dive
The Structure of the Inference Bottleneck — Why Now Is the Turning Point
Memory Wall x Inference Cost: Why the HBM Supercycle Is a Structural Shift, Not a Cyclical Peak