Memory Chips Enter a Super Cycle; Bank of America Significantly Raises Target Prices Across the Supply Chain

Overview

This weekly report from Bank of America expresses an extremely optimistic view on the global memory chip industry. It notes that recent data—including April revenues from Taiwanese memory companies (e.g., Nanya Technology, Phison Electronics) and South Korean semiconductor export figures—have shown triple-digit year-over-year growth, far surpassing market expectations. The core drivers are robust increases in average selling prices (ASPs) for DRAM and NAND chips and strong long-term demand fueled by AI. Based on this, Bank of America has substantially raised its 2026–2028 earnings forecasts for seven covered memory supply chain companies, including Samsung Electronics and SK Hynix, and correspondingly lifted target prices, believing most stocks still have 30–50% upside even after significant recent gains.

Core views

On the demand side, AI is the primary catalyst. The report highlights that capital expenditures by the U.S. Big Five tech giants (Amazon, Microsoft, Google, Meta, Oracle)—which grew strongly by 65% in 2025—are expected to increase by over 60% again in 2026, reaching $600 billion, and exceed $750 billion by 2027. These massive investments are primarily directed toward AI servers and data centers, directly boosting demand for high-bandwidth memory (HBM) and high-capacity storage. On the supply side, capacity allocation toward high-margin products has tightened supply for legacy products. Chipmakers are aggressively shifting production capacity to HBM and server-grade DRAM, constraining supply of traditional DRAM (e.g., DDR4), which has become a key factor pushing its prices to historic highs. The report notes that current spot prices for 16Gb DDR4 are around $60, compared to a previous peak of only ~$10 in 2017. On pricing, memory chips are in a historically strong upcycle. Bank of America’s proprietary memory industry indicator has soared above 180, far exceeding its historical peak (~140), indicating unprecedented industry strength. Although some spot prices have recently pulled back slightly, overall levels remain well above normal. The report specifically mentions that DDR5 prices have remained stable recently, while DDR4 has corrected after its earlier sharp rally. At the company level, earnings upgrades are the core reason for higher target prices. Bank of America has broadly raised its 2026–2028 EPS forecasts by approximately 10%. For chipmakers (Samsung, Hynix, Nanya), this is primarily based on higher ASP and sales volume assumptions; for equipment makers (e.g., Hanmi Semiconductor, ASMPT), it is more driven by strong order expectations from new wafer fabs coming online in 2028.

Analysis framework

The report employs a classic top-down and bottom-up integrated analytical framework. First, it starts at the macro/meso level by tracking high-frequency data (e.g., monthly South Korean semiconductor export reports, monthly revenue from Taiwanese firms) and its proprietary Bank of America memory indicator to assess the current phase and strength of the industry cycle. The report argues the industry is in an AI-driven upcycle of exceptional strength, surpassing any previous cycle. Next, based on this industry outlook, it translates macro tailwinds into company fundamentals. The team models how stronger chip prices and volumes impact each company’s revenue, gross margin, and ultimately EPS. The report details revised key assumptions for each company’s 2026–2028 ASP, shipment volumes, margins, etc., and shows comparisons between old and new models. Finally, it moves to valuation. Analysts applied appropriate valuation methodologies (primarily P/E-based) for each company, assigning different target P/E multiples based on their industry position, growth stage, and business characteristics. For example, for SK Hynix—a leader with high earnings visibility—the target P/E was raised but remains well below its historical average, suggesting further re-rating potential. For equipment makers, forward earnings corresponding to their benefit timing from the capex cycle (e.g., 2028) were used for valuation.

Methodology notes

• Industry/Supply Chain Analysis FrameworkVolume-price decomposition

  Revenue drivers in the memory chip industry can be broken down into bit growth (volume) and average selling price (ASP).

  When analyzing memory chip manufacturers (e.g., Samsung, Hynix), Bank of America explicitly decomposes revenue growth into DRAM/NAND bit growth and ASP changes. By forecasting these two core variables separately, they derive future revenue and profitability.

• Cycle & Sentiment FrameworkInflection Point Analysis

  Constructing a composite indicator to identify inflection points and strength of industry cycles.

  Bank of America uses its proprietary 'Bank of America Memory Indicator,' which integrates multiple dimensions including DRAM/NAND prices, Korean exports, ASPs, and shipment volumes. A reading above 100 typically signals an upturn in sentiment. The report notes the indicator reached a record high of 189, far exceeding prior peaks (~140), serving as key evidence for the current super-cycle.

• Valuation MethodologyPE/PEG valuation

  Price-to-Earnings (P/E) valuation, the ratio of share price to earnings per share, is a core metric for assessing stock valuation levels.

  This is the core method used in the report to value all seven companies and set target prices. Analysts assigned each company a 'fair' forward P/E multiple based on 1–3 year earnings expectations, then multiplied it by the corresponding EPS forecast to derive the target price. The report thoroughly justifies why different P/E multiples were assigned and compares them to historical averages and industry cycle phases.

• Industry/Supply Chain Analysis FrameworkSupply-demand framework

  Commodity prices are determined by the relative balance of supply and demand.

  The core analytical logic: AI-driven demand (demand side) is growing rapidly, while chipmakers reallocating capacity to HBM has constrained supply of legacy DRAM (supply side), creating an imbalance that has driven historic ASP increases for memory chips.

• Industry/Supply Chain Analysis FrameworkUpstream-Midstream-Downstream Transmission

  Analyzing how industry sentiment transmits from upstream chip manufacturing to midstream equipment/materials and downstream applications.

  The report covers not only chipmakers like Samsung and Hynix but also equipment suppliers like Hanmi Semiconductor and ASMPT. The logic is: AI demand → increased chip demand → chipmakers’ profits surge and capex rises → capex converts into new orders for semiconductor equipment, benefiting equipment makers.

• Company Fundamentals & Financial FrameworkROIC–WACC spread

  Return on Equity (ROE), measuring a company’s efficiency in generating profit from shareholders’ equity.

  When valuing Phison Electronics, analysts specifically noted its forecasted 2026 ROE would exceed 50%, far above its historical range of 10–30%. This exceptionally high return on capital is a key argument supporting its premium valuation above historical P/E averages.

Asset mapping & comparison

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

• Samsung Electronics (005930 KS)

  As the world’s largest memory chipmaker, it directly benefits from soaring DRAM/NAND prices and AI-driven HBM demand.

  Strengths

  Optimistic 1Q26 results; highly positive on HBM4 outlook; significant exposure to legacy DRAM.

  Weaknesses

  Report notes limited sequential operating profit growth expected after 2Q26, plus risks from U.S. tariffs and foundry losses.

  Comparison

  Target P/E (10x) is lower than Hynix (8x); larger market cap but relatively less growth elasticity.

  Risks

  U.S. tariff risk, execution risk on 12/16-layer HBM, memory price declines, massive foundry losses, aggressive capex.

• SK Hynix (000660 KS)

  Dominant in HBM3e/4 markets, one of the biggest beneficiaries of AI server memory upgrades.

  Strengths

  Strong 1Q26 results; solid HBM leadership; very high margins.

  Comparison

  Report argues its proven strong performance and AI leadership warrant P/E re-rating from current 8x toward historical average of 12x.

  Risks

  More aggressive capacity expansion by competitors and uncertainty in demand from U.S. tech giants.

• Nanya Technology (2408 TT)

  Benefiting from the current DRAM super-cycle, with April revenue hitting record highs and strong ASP increases.

  Strengths

  Record revenues in Jan–Apr 2026, with YoY growth of 500–700%.

  Weaknesses

  Weaker technology and HBM positioning compared to Samsung and Hynix.

  Comparison

  Target P/E (8x) is below historical average (13x) but above prior cycle peak (6x); report sees potential for AI-driven re-rating.

  Risks

  Rapid transition from DDR4 to DDR5 weakening legacy demand, and competitive pressure from Chinese memory capacity expansion.

• Phison Electronics (8299 TT)

  As a NAND solution provider, benefits from firm NAND pricing and surging demand for AI ecosystem modules (eSSD, aiDAPTIV, etc.).

  Strengths

  Record 1Q26 operating margin (36%); AI product mix jumped from <10% to 38%; management expects sustained high margins.

  Weaknesses

  P/E valuation is above historical average, partly dependent on continued high growth.

  Comparison

  Stronger earnings momentum than other equipment makers, but P/E multiple has been revised down, reflecting market concerns about cyclicality.

• Hanmi Semiconductor (042700 KS)

  Key beneficiary of HBM capacity expansion, especially from large-scale TCB equipment orders tied to SK Hynix’s Yongin and Samsung’s Pyeongtaek P5 fabs starting in 2028.

  Strengths

  Strong position in TCB (thermo-compression bonding) equipment; direct beneficiary of HBM capacity ramp.

  Comparison

  Target P/E (47x) is much higher than Wonik IPS (23x), reflecting its unique position and higher growth certainty in HBM equipment.

  Risks

  Risk of market share erosion from competitors (e.g., ASMPT, Hanwha Semitech).

• ASMPT (522 HK)

  Also benefits from strong demand for TCB (thermo-compression bonding) equipment from both memory and logic chipmakers, especially with long-cycle order growth in 2027–2028.

  Strengths

  Expected EPS uptrend in 2027–28; company optimistic on TCB order outlook.

  Weaknesses

  Faces intense competition from Korean and European rivals.

  Comparison

  Valuation (35x) is above long-term historical average (25x) but below the >100x peak seen in 2024–25 due to abnormally low EPS.

  Risks

  Market share loss to Korean/European competitors.

• Wonik IPS (240810 KS)

  As a CVD equipment supplier, benefits from increased capex by Samsung and other memory makers, but competes globally.

  Strengths

  Report sees earnings inflection point as memory makers increase capex.

  Weaknesses

  Must compete against global leaders like Applied Materials, LAM Research, and Tokyo Electron; relatively weaker moat.

  Comparison

  Only company rated 'Underperform'; P/E (23x) is lower than Hanmi and ASMPT, reflecting competitive disadvantages.

  Risks

  Significant market share erosion by foreign equipment makers or sharp reduction in orders from Samsung/Hynix.

Key data

• Nanya Technology April Revenue YoY Growth+717%Far exceeded expectations and was a key trigger for the earnings forecast upgrade.
• Phison Electronics April Revenue YoY Growth+237%Also far exceeded expectations, indicating strong NAND ecosystem demand.
• South Korea Semiconductor Exports YoY Growth+173%Macro data validates strong micro-level company performance.
• DRAM/NAND ASP QoQ Increase+40–50%Far exceeded market consensus of ~30%, directly driving earnings beats.
• Bank of America Memory Indicator (Feb/Mar 2026)186/189Record high, far surpassing the prior cycle peak (~140), indicating exceptional industry strength.
• Phison Electronics 1Q26 Operating Margin36%Record high, primarily driven by strong contributions from AI ecosystem solutions.
• U.S. Top 5 Tech Giants’ 2026 Capex ForecastGrowing over 60% YoY to $600 billionAI-driven demand is the core source of long-term confidence.
• Samsung Electronics Target Price Increase+19%New target price W370,000, based on 10x 2026–27E P/E.
• SK Hynix Target Price Increase+17%New target price W2,100,000, based on 8x 2026–27E P/E.

Impact & implications

The report argues the memory industry is in an unprecedented AI-driven super-cycle. In the short term, strong chip prices will directly translate into record earnings for related companies. In the long term, continuous AI model iteration and massive capex from hyperscalers provide high certainty for demand growth through 2027–2028 and beyond. This implies: memory chipmakers (Samsung, Hynix, Nanya) will enjoy a ‘double boost’ from rising volumes and prices, with profitability likely peaking in 2026; memory solution providers (Phison) will benefit from higher-value AI product mix, driving structural margin improvement; and semiconductor equipment makers (Hanmi, ASMPT, Wonik) will enter a new capex cycle driven by new wafer fabs, with order visibility extending into 2028.

Risks

• Memory chip prices, after surging, could decline faster than expected, especially for legacy products like DDR4.
• If AI-related capex falls short of expectations, growth in high-end memory demand (e.g., HBM) could slow.
• Samsung faces U.S. tariff risks, execution risks on 12/16-layer HBM, and ongoing foundry losses.
• More successful capacity expansion by Chinese memory makers could create new supply shocks.
• Equipment makers like Wonik IPS face intense competition from global giants (Applied Materials, TEL, etc.), casting doubt on market share sustainability.

What to watch

• Monitor subsequent trends in DRAM/NAND spot and contract prices, especially whether DDR5 remains stable and if the DDR4 correction ends.
• Track actual quarterly capex from the U.S. Top 5 tech giants to validate AI demand assumptions.
• Watch progress on new wafer fab construction and capex plans by Samsung, Hynix, Micron, etc., for 2027–2028.
• Monitor whether Phison Electronics can sustain or further increase revenue contribution from AI ecosystem modules (eSSD, aiDAPTIV, etc.).
• Track order wins by equipment makers like Wonik IPS to verify their ability to gain share amid competition.