Japan IT Hardware & Electronics: Bernstein initiates Japan IT Hardware & Electronics positively on a Physical AI-driven infrastructure and earnings cycle.

Overview

Bernstein launches coverage of seven Japanese IT hardware and electronics companies with a positive sector view. Its central thesis is that Physical AI broadens AI spending into real-world infrastructure and favors firms that can translate installed assets, technical moats and services into durable earnings growth.

Core views

Bernstein defines Physical AI as AI that perceives, reasons, acts and learns in the physical world. This shifts the AI value chain beyond models, cloud and compute into power infrastructure, control systems, operational technology, components, cybersecurity and systems integration. The report argues that Japanese industrial and electronics companies are well placed because their existing strengths lie in high-reliability hardware, installed bases, engineering know-how and mission-critical systems. The key test is not simply AI exposure, but whether it converts into backlog, pricing power, recurring revenue, margin expansion, ROIC and free-cash-flow durability. The first earnings channel is a broader infrastructure capex cycle. Bernstein forecasts data-center power demand to grow at a 19% CAGR through 2032E, making power delivery, conditioning and stability increasingly important. This supports grid equipment, UPS, energy storage, battery backup units and premium electronic components. Server MLCC demand is forecast to grow at a 22% CAGR through 2032E, with total data-center MLCC demand rising from 48bn units in 2024 to 234bn units by 2032E. Newer accelerators may require up to 50,000 MLCCs versus 5,000 in conventional servers. Bernstein sees Murata as the clearest covered beneficiary because premium MLCC supply is concentrated and its technical capabilities support high-end share, mix and pricing. Power demand also drives the energy thesis. Global electricity demand is projected to rise from 29,800TWh in 2023 to more than 40,000TWh by 2035E, while data centers could account for up to 5% of consumption by 2035E from 1% currently. Bernstein identifies gas turbines, grid renewal and nuclear generation as the main demand channels. It prefers MHI in GTCC because the market is highly concentrated, GTCC efficiency reaches about 65% versus 40% for simple turbines, and new installations create multi-year aftermarket revenue. Hitachi and Mitsubishi Electric are favored in high-voltage transformers, switchgear, substations and grid controls, where customization, qualification requirements and long lead times support margins and revenue visibility. Defense is another Physical AI growth driver. The report expects Japan's defense budget to rise from JPY10.6Tn in 2026E to JPY24.8Tn in 2032E, a 15% CAGR. It argues that value will increasingly accrue to systems integrating sensors, communications, secure networks, AI processing and lifecycle services rather than to standalone platforms. MHI is presented as the broadest defense beneficiary, with leadership across aircraft, missiles, naval assets and integrated systems. Mitsubishi Electric has defense-electronics and radar exposure, while NEC benefits from cyber, C4I, networks and communications. Potential overseas defense contracts of roughly JPY1.6Tn provide additional optionality, with MHI estimated to account for 48% of the opportunity set. The second earnings channel is Japan's transition from DX to AX. Bernstein argues that Physical AI needs digitized assets, connected workflows, data platforms, OT-IT integration and cybersecurity before real-world AI can scale. Japan's lower IT spending intensity, lower AI adoption and lower in-house software share than the US are presented as evidence of a multi-year modernization opportunity. This favors domestic integrators with existing customer relationships and mission-critical implementation capabilities. Hitachi's opportunity is linked to Lumada and its installed industrial, rail and energy assets; NEC's BluStellar is expected to rise from JPY542bn in 2025 to JPY1,515bn in 2031E, with scenario and offerings revenue increasing from about 15% to 44% of the mix. The third channel is business-model transformation. Bernstein argues that Physical AI raises the value of software, services, monitoring, maintenance and lifecycle support around installed hardware. The proposed flywheel is installed assets generating data, AI using that data to improve performance, and services monetizing the improvement through recurring revenue and stronger customer lock-in. Hitachi targets Lumada revenue mix to rise from 38% in FY2025 to 50% in FY2027E. MHI's aftermarket share is expected to increase from 25.5% in 2024 to 35.6% by 2031E. The report sees Hitachi, MHI, NEC and Mitsubishi Electric as the strongest transformation stories, while Murata already occupies a high-margin component bottleneck. Bernstein's MAPS framework ranks companies on Market Opportunity, Advantage, People and Scorecard, each weighted 25%. Hitachi and MHI rank highest, with NEC also screening well. Hitachi is Bernstein's highest-conviction name: it combines grid exposure, Lumada-led installed-base monetization and portfolio improvement. MHI combines GTCC scarcity, defense leadership and aftermarket expansion. Mitsubishi Electric offers broad exposure to energy, defense and data-center infrastructure plus Serendie-led solutions. NEC offers AX and defense-network monetization. Murata is the premium MLCC beneficiary. KHI remains a secondary energy and defense player with lower cash-flow conversion, while Panasonic has meaningful BBU and Blue Yonder opportunities but faces share-retention, execution and legacy-margin concerns. Valuation uses SOTP as the primary framework for the conglomerates, triangulated with DCF and two-year blended forward EV/EBITDA; Murata is valued with a 30x P/E on two-year blended forward EPS, triangulated with DCF. Bernstein sets target prices of ¥7,300 for Hitachi, ¥7,000 for Mitsubishi Electric, ¥6,300 for MHI, ¥5,800 for NEC, ¥10,000 for Murata, ¥2,500 for KHI and ¥4,500 for Panasonic.

Analysis framework

Bernstein first maps Physical AI into data centers, energy, defense, DX/AX and portfolio transformation. It then assesses each company using its MAPS scorecard for market opportunity, technology advantage, management execution and financial delivery. Finally, it values conglomerates through SOTP, cross-checks with DCF and forward EV/EBITDA multiples, and uses forward P/E plus DCF for Murata.

Methodology notes

• Competition & strategy

  MAPS framework

  Bernstein scores Market Opportunity, Advantage, People and Scorecard equally to compare structural exposure, technology position, execution and financial delivery across the coverage universe.

• Valuation methodsSOTP (Sum-of-the-Parts) Valuation

  Sum-of-the-parts valuation

  The report values the conglomerates by assigning segment-specific earnings assumptions and peer-based EV/EBITDA multiples, then adds net cash and applies relevant discounts.

• Valuation methodsDCF (Discounted Cash Flow)

  DCF cross-check

  Discounted cash flow is used to triangulate target prices, using company-specific WACC assumptions and a 3% perpetual growth rate.

• Valuation methodsEV/EBITDA valuation

  Two-year blended forward EV/EBITDA

  The report compares target valuations with historical trading ranges, peer multiples and estimates based on two-year blended forward EBITDA.

• Valuation methodsP/E and PEG Valuation

  Forward P/E for Murata

  Murata is valued at 30x two-year blended forward EPS, reflecting its expected premium-MLCC growth, margin expansion and AI-infrastructure positioning.

Asset mapping & comparison

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

• Hitachi (6501.JP)

  Highest-conviction installed-base monetization and grid-infrastructure exposure.

  Strengths

  Grid bottlenecks, Energy backlog, Lumada, OT-IT integration and portfolio optimization.

  Comparison

  Highest MAPS profile and the cleanest Physical AI platform in coverage.

  Risks

  Energy slowdown, weaker HMAX/Lumada adoption and competition in Rail and Connective Industries.

• Mitsubishi Electric (6503.JP)

  Infrastructure-led Physical AI exposure across defense, energy, data centers and solutions.

  Strengths

  Grid systems, defense electronics, factory automation and Serendie platform.

  Weaknesses

  Margin transformation still requires execution.

  Comparison

  Broader end-to-end infrastructure exposure than component-focused names.

  Risks

  Slower data-center or grid investment, mature-business competitiveness and lower-than-expected defense allocation.

• Mitsubishi Heavy Industries (7011.JP)

  Preferred energy-generation and defense platform.

  Strengths

  GTCC leadership, defense-prime breadth, installed base and aftermarket growth.

  Comparison

  Preferred over KHI due to large-frame GTCC leadership and broader defense positioning.

  Risks

  GTCC overcapacity, weaker energy investment, lower defense allocation and inability to meet demand.

• NEC (6701.JP)

  AX, cybersecurity and defense-network beneficiary.

  Strengths

  BluStellar, mission-critical IT integration, cyber/C4I and defense networks.

  Weaknesses

  Traditional IT-services revenue is expected to decline as AX cannibalizes legacy work.

  Comparison

  The report's strongest software-led Physical AI contender.

  Risks

  Slower AX spending, greater-than-expected base-business cannibalization and lower defense allocation.

• Murata (6981.JP)

  Direct premium-MLCC beneficiary of AI server power density.

  Strengths

  Premium MLCC technology, miniaturization, qualification moat and high-end market share.

  Weaknesses

  Exposure remains sensitive to the data-center demand cycle.

  Comparison

  The most direct component beneficiary, versus systems and infrastructure providers.

  Risks

  Data-center MLCC slowdown, insufficient capacity expansion and peer technology catch-up or price competition.

• Kawasaki Heavy Industries (7012.JP)

  Secondary defense and energy exposure with longer-term robotics optionality.

  Strengths

  Defense backlog, aircraft and naval exposure, and robotics position.

  Weaknesses

  Secondary turbine position, lower cash-flow conversion and low-margin motorsports and rolling-stock exposure.

  Comparison

  Trades at a discount to MHI, which Bernstein views as justified by weaker economics and positioning.

  Risks

  Lower defense budgets, intensified powersports competition and tariffs.

• Panasonic Holdings (6752.JP)

  AI data-center BBU and materials exposure, offset by legacy-business drag.

  Strengths

  Current BBU leadership, data-center battery opportunity and Blue Yonder software asset.

  Weaknesses

  Legacy appliances, lower margins and execution concerns.

  Comparison

  Less certain market-share durability than preferred infrastructure and component names.

  Risks

  Faster BBU/CBU share loss and continued weakness in low-margin legacy businesses.

Key data

• Data-center power demand growth19% CAGR through 2032EBernstein's estimate for AI-server-driven power demand.
• Server MLCC demand growth22% CAGR through 2032EDriven by greater component content in more power-dense AI servers.
• Data-center MLCC demand48bn units in 2024 to 234bn units by 2032EIncludes AI-server unit growth and rising MLCC content per server.
• Global electricity demand29,800TWh in 2023 to more than 40,000TWh by 2035EData centers, electrification and other new loads drive incremental demand.
• Japan defense budgetJPY10.6Tn in 2026E to JPY24.8Tn in 2032EEquivalent to a 15% CAGR in Bernstein's forecast.
• Hitachi Physical AI software TAMUp to JPY35TnAcross energy, mobility and industry applications.
• MHI GTCC order share36% in 2025Ahead of GE Vernova at 33% and Siemens Energy at 28%, according to the report.
• NEC BluStellar revenueJPY542bn in 2025 to JPY1,515bn in 2031EGrowth is tied to standardized AX scenarios and services.
• Murata premium MLCC shareApproximately 55%Bernstein identifies high-end qualified capacity as the key AI-data-center bottleneck.

Impact & implications

The report expects AI monetization to broaden toward Japanese suppliers of grid equipment, power systems, defense architecture, IT integration and premium components. Its preferred companies are those that can combine structural demand with technology barriers and recurring software, services or aftermarket revenue rather than merely selling more hardware.

Risks

• A slower AI data-center buildout or weaker grid investment could reduce demand for energy infrastructure, power equipment, MLCCs and BBU systems.
• GTCC orders could normalize more sharply than expected, capacity could fail to meet demand, or energy investment could weaken.
• Japan's defense budget allocation could be lower than expected for MHI, Mitsubishi Electric, KHI or NEC.
• AX adoption could slow, and BluStellar or Lumada may not deliver the expected revenue and margin improvement.
• Premium MLCC supply expansion, peer technology catch-up or price competition could weaken Murata's earnings case.
• Panasonic may lose BBU/CBU share and fail to offset legacy-business margin pressure through restructuring and new growth businesses.

What to watch

• Data-center power growth, grid-equipment orders and the pace of global energy infrastructure investment.
• GTCC order trends, MHI capacity expansion, installation delivery and the growth of aftermarket revenue.
• Japan's defense-budget execution, procurement mix and the conversion of potential export opportunities.
• Hitachi Lumada/HMAX adoption, Mitsubishi Electric Serendie solution revenue and NEC BluStellar AX growth.
• Premium MLCC demand, qualified capacity additions, ASPs and Murata's mix of AI data-center products.
• Panasonic's BBU market-share retention, battery-line conversion and progress in reducing legacy-business drag.