Bernstein: Japan AI Hardware Valuations Pull Back, Murata Now the Best Entry Point
nashnova research
Bernstein says Japan's AI hardware stocks have de-rated enough to create an entry window — Murata is the top pick now; FY26 earnings estimates are up 2%–11% even as sentiment turned excessively bearish.
Why are AI hardware stocks actually the buy here?
Over the past quarter, Japan's software and IT-services stocks rallied about 30%, while AI hardware names sold off on inflated expectations.
Yet FY26 earnings estimates rose 2%–11% over the same period — fundamentals improved while prices fell. This means → the de-rating reflects overshoot in sentiment, not deteriorating businesses.
In plain terms = profits trending up, stock prices trending down. Bernstein sees that gap as a buying opportunity.
Who does Bernstein like most?
Murata is flagged as the best entry point right now; Hitachi and Mitsubishi Heavy Industries are the long-term top picks.
NEC and Mitsubishi Electric also carry outperform ratings — five names spanning software, components, and energy-defence.
This reflects a barbell strategy across three themes rather than a single-track bet.
Theme one: software and cybersecurity — where is the growth?
Japan's domestic software investment growth is forecast to accelerate from +7.5% in 2025 to +10.3% in 2026. Investor days at Fujitsu and Nomura Research confirmed strong IT-modernisation and AI demand.
A wave of cyberattacks on Japanese companies has raised urgency around security spending.
This means → this theme is driven by domestic demand acceleration, not the global AI narrative — making it relatively higher-conviction.
Theme two: MLCC components — will price hikes stick?
MLCCs — multi-layer ceramic capacitors, a basic electronic component used in huge volumes in phones and servers — are running tight. The book-to-bill ratio sits at 1.5; Japan's July–August MLCC exports grew 30% year-on-year.
Yageo, Samsung Electro-Mechanics, and Taiyo Yuden have already pushed through 10%–20% price increases. Whether those hikes hold broadly is the key catalyst.
Bernstein also flags that BBUs (battery backup units) and CBUs (capacitor backup units) add structural demand from AI servers, but new entrants are intensifying competition — the volume uplift is real, the moat less certain.
Theme three: energy, defence, and the yen — what is driving it?
Grid-equipment consensus estimates point to 25%–30% year-on-year growth. Global gas-turbine demand is projected to rise from 91 GW in 2025 to 100–120 GW, suggesting turbine orders could beat expectations.
Defence-sector revenue is expected to sustain roughly 20% growth, while the dollar-yen move from 145 to 159 this quarter gives exporters a tailwind.
In plain terms = energy and defence supply the fundamental case; yen weakness layers on a currency bonus — but whether the yen keeps weakening is itself a risk variable.
What has to go right for this thesis to work?
Bernstein itself identifies two core verification points: whether MLCC price hikes land on schedule, and whether the yen trend holds.
The first determines earnings leverage for the components theme; the second determines how long the currency tailwind lasts for energy-defence names.
This means → the three themes have independent investment logic, but each carries a "discount-if-not-delivered" condition — this is not unconditional bullishness.
市场有风险，内容仅供研究参考，不构成投资建议。