Bernstein Rates Kioxia 'Underperform', Target Price 17,000 Yen
AI summary card
Bernstein Rates Kioxia 'Underperform', Target Price 17,000 Yen
Kioxia raised NAND demand growth guidance to 22% at Investor Day, but Bernstein believes the company's capex and long-term agreement details are insufficient, remaining cautious on cyclicality and China competition risks, maintaining 'Underperform' rating.
- Kioxia raised industry bit demand growth expectation from 20% to 22%, mainly driven by AI inference eSSD
- Company plans to increase average annual capex to 470 billion yen for FY26-FY28, but below market expectation of 520 billion yen
- Long-term agreements (LTA) cover about 50% of shipments, but few details disclosed, making it difficult to quantify earnings sustainability
- Target FY28 eSSD revenue share over 60% (FY25 approx 35%)
- Expect to start dividend distribution as early as 2HFY26
- Bernstein maintains Kioxia 'Underperform' rating, target price 17,000 yen
Report interpretation
Overview
This report comments on Kioxia's (KIOXIA) Investor Day presentation. Kioxia raised the bit demand growth expectation for the NAND flash industry for the coming years and announced higher capex and R&D investment plans to support AI demand growth. However, while welcoming the upward revision in demand expectations, Bernstein expressed concerns about the company's slightly conservative capex planning, lack of Long-Term Agreement (LTA) details, and the industry's inherent cyclicality and competitive threats from China, thus maintaining an 'Underperform' rating on Kioxia with a target price of 17,000 yen.
Core views
Demand and Strategic Adjustment: Kioxia raised the industry bit demand CAGR from CY25 to CY28 from the previous 20% to 22%. Growth mainly stems from Enterprise SSDs (eSSD), especially demand from AI inference applications, while PC and mobile growth is expected to slow. To this end, Kioxia plans to increase its eSSD revenue contribution to over 60% of total revenue by FY28 (approx 35% in FY25). Capacity and Investment: The company plans to increase average annual capex from FY26 to FY28 to 470 billion yen, and R&D investment to 230 billion yen. However, the 470 billion yen capex plan is below the current market consensus expectation of about 520 billion yen. Finance and Shareholder Return: Despite improved prospects, Kioxia did not raise its previously set mid-range operating margin target of 20%+. The company is considering starting dividend distribution as early as the second half of FY26, but cash will also be reserved for potential M&A and other inorganic growth investments, so Bernstein expects Kioxia will not return all 'excess' free cash flow. Technology and Products: Next-generation BiCS10 technology (based on 332 layers) is under development, expected to start sampling this summer, with mass production one year later if qualified. Kioxia claims its BiCS10 is superior to competitors' 400+ layer products in cost and energy efficiency. Risks and Concerns: Bernstein agrees NAND short-term momentum is solid, but remains alert to industry cyclicality and competitive threats from China. Meanwhile, Long-Term Agreements (LTA) currently cover about 50% of bit shipments, but specific details (such as prepayments/financial guarantees) are rarely disclosed, making it difficult to quantify earnings sustainability and valuation.
Analysis framework
Bernstein's analysis mainly revolves around comparing and evaluating Kioxia's disclosed strategic planning with industry third-party forecasts (such as TechInsights). First, by breaking down Kioxia's raised demand growth (from 20% to 22% CAGR), analyze the structural change of its growth drive shifting from traditional consumer electronics to AI data centers, and evaluate whether the company's corresponding capex and product roadmap can match this trend. Second, by comparing the company's capex plan (470 billion yen/year) with market consensus (approx 520 billion yen), judge whether its capacity expansion strategy is aggressive or conservative, which may affect its future market share and profitability. Finally, focus on key factors for business visibility and earnings quality—the coverage and term details of Long-Term Agreements (LTA). Due to insufficient LTA information, institutions find it difficult to make more precise judgments on the stability of company earnings and valuation, which constitutes an important basis for its cautious view.
Methodology notes
The report judges market prospects and price trends by analyzing industry supply and demand relationships
This report combines TechInsights forecasts cited by Kioxia with the company's own outlook to assess whether the NAND market's supply and demand tightness can sustain until CY27, which is the core method for judging the memory industry prosperity cycle and company profitability.
The report uses Price-to-Book (PB) ratio as the main valuation method for Kioxia and other storage companies
In the valuation method section, Bernstein explicitly mentions using a 2-year forward Price-to-Book ratio (1x) for valuing Kioxia, and derives the target price based on this. This is a common relative valuation method used in the capital-intensive, highly cyclical semiconductor industry.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- Kioxia Holdings (285A.T)Direct analysis object, Investor Day information announcing entity, rated 'Underperform'.
- Strengths
- Demand expectation revised up, clear growth plan and next-generation technology roadmap (BiCS10) in AI eSSD field, considering dividend distribution to improve shareholder return.
- Weaknesses
- Capex plan below market expectation, insufficient LTA details leading to low earnings visibility, did not raise margin target, facing industry cyclicality and China competition risks.
- Comparison
- The report lists target prices and valuation multiples of other storage companies such as Samsung, SK Hynix, Micron in the valuation section, but does not expand on detailed horizontal comparison.
- Risks
- The report clearly points out risks include: demand growth weaker than expected or supply increase leading to favorable pricing environment ending early, investor sentiment affecting valuation, China's progress in NAND field constitutes downside risk.
Key data
- NAND Industry Bit Demand CAGR (CY25-CY28)22%Kioxia latest forecast, revised up from previous 20%, mainly driven by AI eSSD demand
- Kioxia FY26-FY28 Average Annual Capex PlanJPY 470BHigher than FY25's 283.7 billion yen, but below market consensus of ~JPY 520B
- Kioxia FY26-FY28 Average Annual R&D Expense PlanJPY 230BSignificant increase compared to FY25's 130 billion yen
- Kioxia Long-Term Agreement (LTA) CoverageApprox 50%Covers current bit shipments, company is seeking to increase coverage
- Target eSSD Revenue Share (FY28)>60%FY25 approx 35%, showing business structure transformation towards high value-added AI data center storage
Impact & implications
The report believes that Kioxia's strategic adjustment reflects the strong pull of AI on high-end storage demand, and the overall industry demand outlook has improved. However, the company's relatively conservative capex may mean its expansion pace lags behind market demand growth or competing peers, affecting long-term market share. In addition, unclear LTA details and existing earnings volatility, coupled with industry cyclicality and China competition risks, these factors together constitute the basis for Bernstein's cautious view on Kioxia.
Risks
- Industry favorable pricing environment may end early due to demand weaker than expected or supply increase
- Changes in investor sentiment may affect the valuation level the company receives
- Technological progress in China's memory (especially NAND) field constitutes competitive threat