Nihon Dengi Co., Ltd. (1723.T): Research Report

Reporting-period check: Nihon Dengi's fiscal year ends March 31 and it reports every quarter. The latest period should be the first half of FY3/2027 (April to September 2026), with results due in early November 2026. They were released on November 4, 2026, so that period anchors this report.

1. What the Company Does

Nihon Dengi installs and maintains the control systems that run the air-conditioning in large buildings. Think of the sensors, valves, controllers and central monitoring screens that decide how much chilled water reaches an office floor, when a hospital ward gets fresh air, or how a data hall stays cool. Nihon Dengi designs that system, installs it, tunes it until it works, and then services and upgrades it for the life of the building. The Japanese term is keiso (instrumentation): measuring, monitoring and controlling the equipment in buildings and factories, such as air-conditioning and production lines.

Origins. The company exists because of a supply chain that was missing in post-war Japan. In 1952 Yamatake Keiki (today's Azbil) signed a capital tie-up with Honeywell, the US controls maker, and began importing and selling air-conditioning control equipment in Japan. Almost no contractors existed to install it, so Shimada Shichiro and the other founding members set up the company, convinced the market would grow. It began in 1959 as Japan's first company dedicated to HVAC instrumentation, handling design, installation, adjustment and maintenance of automatic air-conditioning control in one place. The incorporation date was September 26, 1959, and the shares listed on March 13, 2003.

That origin still defines the business. Nihon Dengi did not become a manufacturer. It became the engineering arm of Azbil's ecosystem: it mainly handles instrumentation for central HVAC in medium-to-large buildings of 10,000 sqm or more, and it is the leading authorized contractor of Azbil, which the company estimates holds about 70% of Japan's market for central-HVAC automatic controls by shipments.

Three later changes shaped what the company is today:

• A second business line. It took its building engineering into factories. Using its HVAC and engineering know-how plus production-management systems from subsidiary Jupiter Advance Systems, it now automates production and conveyor lines. That is the Industrial Systems segment (Section 2).
• A tilt towards existing buildings. Retrofit and renewal work now earns more of the company's revenue than new construction (Section 2).
• Becoming a training organisation. Skilled engineers are the binding constraint (Sections 7 and 8), so the company set up the "Dengi Academy" in April 2024 to train new and junior engineers, and in 2025 bought land for a dedicated training and R&D campus.

The problem it solves. Central HVAC is the largest energy load in a big building and the hardest system to get right. A tower's chillers, air handlers, pumps and dampers have to respond to occupancy, weather and tenant changes, or the building wastes energy and tenants complain. Owners and contractors need someone who can turn a box of Azbil controllers into a commissioned, energy-optimised system, and who will still be there in 15 years when it needs replacing. Within an HVAC instrumentation market dominated by the Azbil group, Nihon Dengi works as an "engineering company" covering design, installation and maintenance.

Why it is hard to copy (the short version; Section 3 has the detail). Instrumentation is learned on the job. Equity research house Fisco notes that as Azbil's largest dealer, Nihon Dengi has a longer history and more experience than its rivals, and in a field described as "experience engineering", that experience tends to show up as a profitability gap. Each building's control logic is bespoke. Commissioning it needs engineers who understand both the mechanical plant and the control software, and Japan is short of such people.

A worked example. A Tokyo redevelopment tower goes up. The general contractor hands the mechanical package to an HVAC sub-contractor, which in turn subcontracts the controls to Nihon Dengi. Fisco describes the sequence: for new buildings, the company handles the air-conditioning control system in one package, from system design and installation through pre-handover test runs and adjustment to operating instructions at handover; it also builds control systems for district heating and cooling plants. After completion it maintains the HVAC equipment, renews it, and improves energy efficiency and operations to cut energy and running costs. The control panels themselves can come from its own plant: the company opened a factory in Okayama that makes control panels and operating panels. About 15 years later, when the building management system is due for renewal, the incumbent installer is the natural choice to rip out and replace it. That cycle is the economic heart of the company (Section 4).

2. Business Segments

Nihon Dengi reports two segments. They share an engineering method but have very different customers and economics. According to the FY3/2026 segment data, HVAC Instrumentation (空調計装関連事業) and Industrial Systems (産業システム関連事業) split revenue roughly 90% / 10%.

2.1 HVAC Instrumentation (~90% of FY3/2026 revenue)

What it does. This segment designs, installs, commissions and maintains automatic control systems for central air-conditioning in non-residential buildings: buildings, factories, hospitals, schools and commercial facilities. It reports two sub-lines that behave very differently.

• New construction (新設). Controls for buildings being built: redevelopment towers, factories, hospitals, data centers. The work arrives through the construction chain, is lumpy, and follows the building cycle.
• Existing buildings (既設). Renewal, energy retrofits and maintenance of installed systems. In FY3/2026, new-build sales fell 14.6% while existing-building sales rose 20.9%, so existing-building work now makes up roughly two-thirds of the segment and about 59% of group revenue. Management describes it as stock-type revenue: the existing-building share of sales reached 58.8%, with system renewal on a roughly 15-year cycle after construction supporting steady orders.

Core capability. Nihon Dengi can take Azbil's continually changing product range and make it work in large, complex buildings. Fisco explains that new-build orders were piling up because Azbil's evolving products and increasingly complex large projects need engineering capability that, among the dealers, only Nihon Dengi has; the rise in large renovation orders as buildings age has the same cause.

Why existing-building work matters more than its share suggests. It earns better margins and the company often contracts directly: existing-building work is more profitable than new construction, and terms are reportedly better again when the company is the prime contractor. The shift towards renewal work is the main structural change in the business over the last three years.

Role in the group. This is the engine. It produces nearly all group profit and receives most of the scarce engineering headcount.

Main rivals (one line; analysis in Section 5): Azbil's own building-systems engineering arm, other Azbil dealers such as Totech and Ohtech, and Johnson Controls in its installed base.

2.2 Industrial Systems (~10% of FY3/2026 revenue)

What it does. This segment mainly does instrumentation for factories and conveyor lines, various automatic-control works, and production-management systems for food factories. Fisco describes it as an "evolved form" of instrumentation engineering: it uses control technology to automate whole factories or production lines and reduce manual labour, setting up and automatically controlling measuring instruments and robots in production and conveyor processes at plants of every size.

Core capability and why it is separate. The engineering method is shared, but the customer (a factory operator), the scope (process and logistics lines plus software) and the vendor-neutral equipment all differ from the Azbil-centred HVAC business. The production-management software comes from the Jupiter Advance Systems subsidiary. Fisco's view is that factories lack the know-how to automate themselves and few firms can help them, so the market is essentially untapped.

Role in the group. This is the growth option, and it has been slow to mature. An earlier Fisco note observed that the division had only recently started and was supplying staff to the HVAC business, so it had not yet reached full activity. Recent results show traction: in FY3/2026, electrical works and production-management software grew, and the segment's carried-forward backlog rose 36.6%. Management's plan names this segment as its "business domain expansion" strategy (Section 7).

Main rivals: fragmented regional system integrators and factory-automation engineering firms. Section 5 covers this briefly.

2.3 Geographic mix

Nihon Dengi sells only in Japan and has never had an export business. Fisco notes that the company has led the industry from a base in Japan's three largest metropolitan areas, working with Azbil, and its branch network grew outward from Tokyo. The corporate history lists a Nagoya office opened in August 1969 (now the Nagoya branch), plus offices in Chiba and Okayama that are now branches. The demand split within Japan matters more than any foreign question. Greater Tokyo redevelopment drives the new-build pipeline, while regional core cities and factory sites (semiconductors, reshored manufacturing) add a second leg. In FY3/2026 the company cited redevelopment in greater Tokyo and regional core cities, and factory construction from manufacturing returning to Japan as the demand drivers.

3. Products and Operations

Nihon Dengi does not sell a product in the usual sense. It sells an engineered, commissioned control system, and then a long tail of service on it. The things that matter to the investment case are the following.

3.1 What gets delivered

Building automation systems for central HVAC. These are mostly built on Azbil hardware and software: sensors (temperature, humidity, pressure, CO2), controllers, actuators and control valves on the chilled- and hot-water loops, and a central monitoring and management system. Fisco notes that because Azbil and its dealers control most of the market, Azbil equipment has in effect become the industry standard. Nihon Dengi adds the engineering: control-logic design, wiring and installation, programming, commissioning, and tuning for energy performance.

District heating and cooling plant controls. These are control systems for the central plants that serve multiple buildings (see the Fisco description in Section 1).

Control panels and monitoring panels. Panels are built in-house at the Okayama factory. The company's business description covers design, installation, adjustment and maintenance of automatic control systems for building and factory automation; design and fabrication of monitoring and control panels; and sales of automatic control equipment. In-house fabrication gives it control over lead times and quality on its own projects.

Industrial controls and software. These are instrumentation and electrical works for production lines, utility automation (chilled and hot water, steam, compressed air), waste-heat recovery and production-management software. Fisco lists electrical instrumentation works, automation of specialised utility systems (chilled/hot water, steam, compressed air), use of waste hot water and waste heat from plants to cut environmental load and running costs, and improvements in safety and in the accuracy and efficiency of sorting work.

Maintenance contracts. These are the recurring layer on the installed base. Management called out new maintenance contracts as a bright spot in the FY3/2026 results briefing (Section 7).

3.2 The delivery process: where the constraint lives

The value chain for a new-build project looks like this:

1. Bid and design. The company estimates control scope from mechanical drawings and designs the control architecture and point list.
2. Procurement and fabrication. It buys Azbil devices through its dealer agreement and builds panels in Okayama.
3. Installation. Field wiring, device mounting and panel installation are done largely through partner subcontractors (kyoryoku gaisha) under Nihon Dengi's site supervision.
4. Programming and commissioning. Nihon Dengi's own engineers program, test and tune the system before handover. This is the skill-intensive step.
5. Handover and operator training.
6. Maintenance and renewal. This continues for the life of the system, with a full renewal roughly every 15 years.

Retrofit work runs the same chain inside an occupied building, usually with tighter windows and more integration with legacy equipment.

The binding constraint is people. Factories and raw materials are not the limit. The supply of licensed engineers and of partner-company field crews caps how much work the company can deliver. Two things tightened this:

• Overtime caps. From April 2024, overtime in construction is capped at 45 hours a month and 360 hours a year in principle, with hard ceilings even under special agreements. The company has said its order guidance reflects this, citing the overtime ceiling regulation that applied to construction from FY2024.
• Partner-company fragility. The company's own strategy page puts "worsening labour shortage", "co-prosperity with partner companies" and compliance with the Construction Business Act and the new subcontracting law at the top of its environment assessment, and describes an "Authorized Service Partner" scheme aimed at building lasting relationships with partner companies and actively supporting their recruitment and training.

Capacity responses (plant-equivalent events):

• Dengi Academy (April 2024). Before the Academy, new graduates normally did one month of training before on-site OJT; with sites extremely busy, the company now trains them intensively for a full year at head office, and plans to extend the programme to partner companies.
• Narashino campus. In June 2025 the company acquired about 1,500 tsubo of land in Narashino, Chiba, and is building a new training and R&D base, with full operation expected in a few years. The May 2026 briefing set the target as a Dengi Academy plus Technology Innovation Center, aiming to start operating in FY2029.
• Small bolt-on acquisitions of installation crews. In April 2024 it acquired all shares of Taito Setsubi (unconsolidated), and in March 2025 merged Taito Setsubi into ND Tech (also unconsolidated). These are small deals, but they show the company buying field capacity when it cannot hire it.

Revenue timing. Revenue is heavily weighted to the fourth quarter. The company notes that completions and handovers cluster in the fourth quarter (January to March), so that quarter's sales are larger. First-half figures therefore say little about the full year (Section 9).

4. Customers

Who buys. There are three distinct buyers, and each wants something different.

1. HVAC and mechanical sub-contractors on new construction. In Japan's layered construction system, the general contractor wins the job and passes specialist works to sub-contractors, which pass work further down to smaller firms. For new buildings, Nihon Dengi usually sits under the mechanical sub-contractor, which in turn sits under the general contractor. That contractor's project and procurement managers decide, and their criteria are a credible schedule, a fixed price and confidence that commissioning will not delay handover. The sales cycle follows the project: controls are usually awarded once the mechanical design is fixed, and delivery runs one to several years. Fisco reports that some Tokyo projects are slipping, with small and mid-size projects delayed by rising material costs and some not completing until close to 2030.

2. Building owners and facility managers on existing buildings. For renewals, energy retrofits and maintenance, the owner's facilities team (or its property-management company) often contracts directly. Their criteria are knowledge of the installed system, the ability to work in an occupied building without disruption, and proven energy savings. Recent wins cover existing-building work at factories and data centers, and earlier, offices and factories.

3. Factory operators (Industrial Systems). Plant engineering and production-management teams buying line automation, utility controls and software. Demand comes from automation and labour-saving investment, and the work is more bespoke.

Named accounts. None of the sources reviewed for this report disclose named customers or customer concentration. Results releases describe demand by building type (offices, factories, data centers, hospitals, hotels, research and public facilities), not by client.

Why customers choose Nihon Dengi

• Azbil-standard competence at scale. The building's controls are almost certainly Azbil, and Nihon Dengi is the dealer with the deepest engineering bench. Fisco states that as the only pure-play specialist with a dedicated engineering division, it has technical strength recognised by itself and others among Azbil's dealers.
• Capacity in a seller's market. With engineers scarce, contractors pick whoever can actually staff the job. Fisco observed that with many enquiries, the company was choosing which orders to take, and because only a limited number of firms including Nihon Dengi could take on the work, its quotes were readily accepted.
• Incumbency. For retrofit, the firm that commissioned the system holds the as-built knowledge.

Switching costs. These are highest in existing buildings. A building management system is a 15-year-plus asset with custom logic, thousands of points and links to the mechanical plant. Replacing the installer means paying someone else to reverse-engineer it while the building stays occupied. Changing the controls brand (from Azbil to Johnson Controls, for example) is even harder because it means a full system replacement. Switching costs are lower on new builds, where each project is bid afresh, though contractors tend to stick with installers who have delivered on time before.

Concentration. The concentration that matters is on the supply side, not the customer side: the company depends on Azbil (Section 8.2). Azbil is also a shareholder (Section 10).

Contract structures and predictability

• Construction contracts are mostly recognised over time. The interim filings split revenue into goods transferred at a point in time and goods transferred over a period of time.
• Order intake and carried-forward backlog are the leading indicators, and both are at records (Section 7).
• Maintenance contracts recur and are growing.
• Q4 concentration makes quarterly revenue noisy, but the backlog gives better-than-usual full-year visibility for a contractor.

5. Competitive Landscape and Moat

5.1 Industry structure

Japanese building air-conditioning splits in two. Individual (packaged) air-conditioning, with units room by room, suits smaller buildings and is led by large air-conditioner makers. Central air-conditioning in large buildings needs controls. That is Nihon Dengi's market, and it is effectively an Azbil standard. The HVAC instrumentation market is said to exceed ¥160 billion (company research), with Azbil and its dealers, including Nihon Dengi, holding about 60-70%.

So the main competitive fight is not between control brands. It is over who gets to engineer and install Azbil systems, and over a minority of projects using other brands.

5.2 Named competitors

Azbil: partner and competitor. Azbil's building-automation business covers building lifecycle support, environmental control systems, instrumentation design and sales, and engineering. It takes many large projects itself and uses dealers to add capacity. Nihon Dengi wins against Azbil's direct arm because it is extra engineering capacity Azbil does not have to employ, and because it is a dedicated specialist. It loses when Azbil decides to keep a flagship project or a key account in-house. The relationship is symbiotic, and Azbil also owns a stake in Nihon Dengi (Section 10).

Totech. Totech is the No.1 sales agent for Daikin, Japan's top packaged-air-conditioner maker, and an Azbil dealer in instrumentation; it describes itself as the only equipment distributor in Japan that can also install instrumentation. Totech wins where a customer wants equipment and controls bundled. Nihon Dengi wins on complex central-HVAC engineering, where its whole organisation does nothing else.

Ohtech. It is the closest pure-play peer: mainly air-conditioning automatic-control works and maintenance, plus sales of piping materials. Azbil also holds 4.39% of Ohtech. Being part of the same dealer family, the two compete on capacity and regional coverage rather than technology. Nihon Dengi's advantage is scale and the depth of its engineering bench.

Johnson Controls (and other global brands). These are the alternatives when an owner or designer specifies a non-Azbil system. In Japan they are a minority because Azbil is the incumbent standard. Global rivals have more software scale, but switching an installed base is costly (Section 4).

Industrial Systems. Competition here is fragmented among factory-automation integrators and electrical contractors, and Nihon Dengi is a small player. There is no moat in this segment yet.

5.3 Market share and why

The company does not publish its own share figure. The structure is clear, though: the Azbil camp holds the majority of a market worth over ¥160 billion, and within that camp Nihon Dengi is consistently described as Azbil's largest dealer. The share is what it is because of history (it was the original installer of Yamatake-Honeywell equipment), because of the installed base it services, and because engineering bench depth scales slowly.

5.4 Barriers to entry

• Skilled labour and licences. A new entrant must hire commissioning engineers who are already scarce industry-wide. This is the highest barrier, and it is getting higher.
• Access to the standard. Being an authorised Azbil dealer is a gate that Azbil controls.
• Installed base. Retrofit work goes to whoever knows the system.
• Track record with contractors. Sub-contractors on ¥100bn-plus towers do not risk handover on an unproven installer.

5.5 Structural shifts

• From new-build to renewal. This favours incumbents with large installed bases.
• Labour scarcity has turned the market into a seller's market. Installers can be selective about orders (Section 4), which lifts margins across the Azbil camp. Peers have prospered too: Ohtech raised its year-end dividend after strong results.
• Digitalisation of building management (cloud BEMS, analytics) could in time move value towards software owners, mainly Azbil and global brands, and away from installers.

Moat: Narrow - switching costs, intangible assets (privileged Azbil channel position and engineering know-how), efficient scale

Section 4 showed that existing-building customers face real switching costs. Section 5 shows that Nihon Dengi holds the deepest engineering position inside a de facto standard that the Azbil camp dominates. The company has also been the leading installer for six decades, through several construction cycles. It does not earn "Wide" because the advantage depends on one supplier's channel policy and on a labour shortage that currently favours incumbents. The company does not own the standard. The thing most likely to erode the moat is Azbil changing how it allocates engineering work: insourcing more projects, or shifting system architecture towards software and remote services that need less local commissioning.

6. Industry

What drives demand

1. The non-residential construction cycle. New-build controls work follows office, hospital, factory and data-center construction. The company described FY3/2026 conditions as a buoyant environment from greater Tokyo and regional core-city redevelopment and factory construction tied to manufacturing returning to Japan, against a backdrop of labour shortages common to the whole construction industry.
2. The ageing building stock. Japan's large stock of 1990s-2000s buildings is reaching its second or third controls renewal. This demand is steadier than new construction and grows with the installed base.
3. Energy-efficiency regulation. Since April 2025, essentially all new homes and non-residential buildings must comply with energy-efficiency standards; buildings that fail are not granted building confirmation and cannot start construction. Further tightening is coming: in April 2026, the BEI standard for mid-size non-residential buildings is raised substantially. Controls are among the cheapest ways to reduce a building's primary energy use, so stricter rules increase both the scope and the sophistication of controls work.
4. Data centers and semiconductors. These are new, controls-heavy building types with critical cooling loads. The whole contracting sector reports them as growth: listed electrical contractor Nippon Densetsu Kogyo, for example, notes continued private demand including data-center related work keeping orders high.

Size and trajectory. HVAC instrumentation is a mid-sized niche, said to exceed ¥160 billion. Older estimates were higher: in 2018 Fisco cited around ¥200 billion, with roughly 80% held by Azbil and its dealers. The published estimates differ, which suggests nobody measures the market precisely. Growth is better tracked through order books than through any industry total, and the order books of all listed participants are rising.

Position in the supply chain. The chain runs from component makers to the controls brand (Azbil) to the engineering dealer/installer (Nihon Dengi) to the mechanical sub-contractor to the general contractor to the building owner. Nihon Dengi captures the engineering and service layer, which is where labour, know-how and installed-base relationships sit.

Import substitution. Not relevant. The market is domestic and the standard itself is a Japanese adaptation of a Honeywell lineage.

Regulation and policy affecting supply

• The construction overtime cap since April 2024 (Section 3) limits industry capacity. It raises costs, but it also restricts supply and so supports pricing.
• Subcontracting fairness law: the company itself cited the impact of compliance with the Torihiki Tekiseika Act (the act on fair transactions with small and medium-sized subcontractors) in setting its FY3/2027 outlook. This pushes value down the chain to partner companies, which raises the cost of subcontracted labour.

Cyclicality. New construction is cyclical with long lags. Orders are taken years before completion, so a downturn shows up in orders well before it shows up in revenue. Renewal and maintenance work is far less cyclical, because owners can delay renewal but cannot defer it forever. The shift in mix towards existing buildings has therefore made the industry less cyclical for incumbents. The structure has not yet been tested through a serious non-residential downturn.

Tailwinds: an ageing stock of buildings, energy regulation, data centers and factory reshoring, a Tokyo redevelopment pipeline running to about 2030, and capacity limits that support pricing. Headwinds: a shrinking skilled workforce, ageing and succession problems at partner companies, construction cost inflation delaying smaller projects, and a long-term decline in the number of new buildings as Japan's population falls. The company's own long-range planning accepts that with population decline, Japan will no longer see new buildings going up one after another, so new-build HVAC work is unlikely to expand much beyond current levels over the long term.

7. Growth Triggers

Ranked by potential impact on the business. Sources are the six results events listed in Section 9.

• Raised long-range targets backed by a record order book. At the FY3/2026 results briefing (May 27, 2026), management raised the final-year (FY2027) targets of the mid-term plan to ¥52bn of sales and ¥13bn of operating profit, and published a financial strategy (cash allocation) running to FY2030. It also lifted the FY2030 ambition to ¥60bn of sales and ¥15bn of operating profit. This is the third upward revision of the same plan: it was also raised in February 2025 (Section 9). Supporting it, the Q1 FY3/2027 release (Jul 24, 2026) reported the HVAC carried-forward backlog up 37.1% year on year, with new-build backlog up 44.1% and existing-building backlog up 31.2%.

• Greater Tokyo redevelopment runs to about 2030. At the May 27, 2026 briefing, management said Tokyo redevelopment projects will continue to around 2030, with several projects already at the enquiry stage. The same theme of large Tokyo and regional redevelopment projects appeared in the Q2 FY3/2026 release (Nov 4, 2025) and the FY3/2026 release (May 7, 2026). It has been repeated.

• Data centers and factories as new demand pools. Data-center work moved from new-build orders (Q2 FY3/2026, Nov 4, 2025: new-build orders rose for offices and data centers) to existing-building orders (Q1 FY3/2027, Jul 24, 2026: new-build orders rose for factories and hotels, existing-building orders for factories and data centers). Factory demand linked to reshoring was cited at every event in the window.

• Renewal cycle and maintenance contracts as a recurring base. The FY3/2026 briefing (May 27, 2026) highlighted the 15-year system renewal cycle after construction and reported that new maintenance contracts came in 65.8% above the internal budget. The mid-term plan's HVAC strategy is labelled "strengthening the existing business", with renewal and energy-saving retrofit at its centre.

• More capacity: Dengi Academy, the Narashino campus and partner-company support. At the May 27, 2026 briefing, management called the labour shortage the biggest bottleneck to growth, making hiring, training and partner-company support the top priority, and set FY2029 as the start date for the Narashino training and innovation center (Section 3). This trigger has been repeated across the plan period. Each increase in engineering capacity converts directly into orders the company can accept.

• A war chest for M&A and human-capital investment. The FY2030 cash-allocation plan (May 27, 2026) set out ¥48bn of operating cash generation to FY2030 plus ¥22bn of distributable funds, allocated to ¥38-43bn of growth investment, ¥7bn of basic infrastructure and ¥20-25bn of shareholder returns, with a clear commitment to M&A and human-capital investment. Past acquisitions have been small (Section 3), so a larger deal would be new for this company.

• Industrial Systems scale-up. The Q1 FY3/2027 release (Jul 24, 2026) reported Industrial Systems orders up 32.0%, with segment backlog up 41.0% and growth in electrical works and software, following similar gains in production-management systems reported at Q2 FY3/2026 (Nov 4, 2025). Management positions the segment as its domain-expansion strategy.

• Unwinding strategic shareholdings to fund growth and returns. With the Q1 FY3/2027 results (Jul 24, 2026), the board resolved to cut strategic shareholdings to below 10% of consolidated net assets by FY3/2031, with the proceeds going to growth investment and shareholder returns.

8. Key Risks

8.1 Engineering capacity caps growth and could squeeze margins. Mechanism: Revenue is limited by how many projects licensed engineers and partner crews can commission (Section 3). If attrition at partner companies (ageing owners, no successors) outpaces Academy output, the company must turn away orders, or accept them and run late, which brings penalties and cost overruns. Subcontracting-law compliance and wage increases raise the cost of each crew. Management itself flags the labour shortage as the bottleneck, so construction capacity sets the ceiling on orders and sales, with generational change at partner companies a further issue. The company also announced a rise in starting salaries for new graduates and better treatment of employees in August 2026. Calibration: high probability, moderate drag. It is already affecting the business and is unlikely to break it.

8.2 Dependence on Azbil. Mechanism: Nihon Dengi's HVAC franchise rests on Azbil's product standard and dealer system (Section 5). Azbil could insource more engineering, reallocate territories or key accounts among dealers, change dealer margins, or move its product architecture towards software and remote services that need less local engineering. Any of these would cut both volume and pricing power, and Nihon Dengi has no comparable second brand to turn to. Calibration: low probability of a sudden break, given six decades of partnership and Azbil's shareholding (Section 10). The impact would be high, and slow erosion is more plausible than a rupture.

8.3 Peaking new-build cycle and project slippage. Mechanism: New-build orders depend on a Tokyo redevelopment wave that management itself dates to about 2030, and on factory and data-center spending that can stop quickly when the cycle turns. Material-cost inflation is already pushing smaller projects back (Section 4). Management also cited Middle East risk when setting FY3/2027 guidance. Calibration: moderate probability over the next three years, moderate impact. The large renewal book cushions the business, but new builds remain about a third of HVAC sales and feed the future renewal base.

8.4 Margin normalisation from a seller's market. Mechanism: Recent profitability reflects selective ordering in a capacity-short market (Section 4). If demand cools, or competing installers add capacity, pricing power fades, and the high-margin projects that boosted recent results will not repeat. Management has already guided down: at the May 2026 briefing the President said the prior year's high margin reflected the completion of highly profitable projects, and the lower margin forecast for this year is a normalisation to usual levels. The first sign appeared in the latest quarter: operating profit in the July-September quarter fell 3.1% year on year on 8.0% higher sales. Calibration: high probability of some normalisation, moderate impact.

8.5 Execution risk on large projects and Q4 timing. Mechanism: Revenue and profit depend on completions in January to March (Section 3). A large tower whose handover slips past March 31, or a fixed-price contract with a commissioning overrun, can swing a full year. Bigger tickets make this worse: management noted rising construction costs have increased the value per project, which concentrates risk in fewer jobs. Calibration: moderate probability in any given year, mostly an earnings-volatility risk rather than a threat to the business model.

8.6 Capital-deployment risk from the new M&A budget. Mechanism: The FY2030 plan earmarks up to ¥43bn for growth investment including M&A (Section 7). That is large relative to a company whose acquisitions have been small installation crews. Buying a weaker installer or an industrial integrator at a cycle peak could dilute returns and absorb management attention, which is already stretched by the capacity problem. Calibration: low-to-moderate probability, with the impact depending on deal size. There is no track record to judge this by.

9. Walk the Talk

The six results events used (Nihon Dengi publishes results releases and hosts results briefings; no transcript was available, so the releases, briefing summaries and guidance-revision notices are the record):

1. Q1 FY3/2026: results release, Jul 25, 2025
2. Q2 FY3/2026: results release and guidance revision, Nov 4, 2025 (briefing video posted Dec 1, 2025)
3. Q3 FY3/2026: results release and guidance/dividend revision, Jan 28, 2026
4. FY3/2026: results release May 7, 2026; results briefing May 27, 2026
5. Q1 FY3/2027: results release, Jul 24, 2026
6. Q2 FY3/2027: results release, Nov 4, 2026 (within 90 days of this report)

The pattern behind the beats. Nihon Dengi's guidance has been conservative by design. Its own FY3/2027 outlook statement says it was set after comprehensively considering Middle East risk and the impact of complying with the Torihiki Tekiseika Act. Analysts read it the same way. Even before the run of upgrades, Fisco wrote that the company tends to set conservative budgets, and that growing capacity from training, better margins won at the proposal stage, or firm-wide balancing of regional supply and demand could lead to upgrades, as in prior years. In the most recent full year this happened twice before year end, and the final result still came in above the last revision.

The one call worth watching: the "normalisation" warning. In May 2026, while publishing record results and raised long-range targets, the President described the coming year's lower margin as normal rather than as a deterioration (Section 8.4). Six months later, the July-September quarter delivered exactly that: profit slightly down on higher revenue, with the first-half total still ahead of plan. Management did not chase the record margin with its guidance, and the numbers then moved the way it said they would. This is the best evidence in the window that its forecasts are honest rather than sandbagged.

"本売出し後も売出人となる金融機関及び事業会社との良好な関係は変わらず、取引関係に影響はございません。" ("Even after this offering, our good relationships with the selling financial institutions and business company remain unchanged, and there is no effect on our business dealings.")

• Company announcement of the secondary offering, Sep 28, 2026 (quoted via kabu.96ut.com)

This statement is not a forecast, but it is a commitment to track. The offering and its ownership effects are covered in Section 10.

Assessment. Over these six periods this management team has consistently under-promised. It guided low, raised twice, beat the raise, and moved long-range targets up three times as reality overtook them. Its one cautious call, on margin normalisation, has been accurate so far. The open question is not credibility but untested ground: none of the six periods included a demand downturn or a large acquisition, so there is no evidence yet of how management guides when conditions turn against it.

10. Ownership, Governance and Shareholder Friendliness

Part 1 - Ownership and control. There is no controlling shareholder. The register is a mix of insiders, a supplier, banks and foreign institutions. At March 31, 2026 the largest holders were the Nihon Dengi Employee Shareholding Association (8.29%), Shimada Ryosuke (6.56%), Nagata Kenji (4.97%), Mizuho Bank (4.28%), Azbil (4.07%), Shimada Shoko (3.05%), Shimada Junko (3.05%) and Custody Bank of Japan trust account (2.91%). Fidelity's FMR is also a substantial holder, with a stake built up to 9.45% per the May 2026 briefing coverage. Several holders share the Shimada surname with the President; the filings reviewed do not state their relationships, so none are assumed here. The register changed materially in autumn 2026. On Sep 28, 2026 the company announced a secondary offering of 3,852,600 shares plus an over-allotment of up to 577,800 shares, with sellers including Mizuho Bank, SMBC and President Shimada. The stated aims were better liquidity, broader investor recognition, and a wider base of partner shareholders supporting medium-to-long-term strategy, and the offering followed the company's review of cross-shareholdings. Afterwards, Shimada Ryosuke's holding fell from 6.36% to 5.68% (3,726,837 shares), with the reporting obligation arising on Oct 6, 2026, and Mizuho Bank's reported holding fell to 2.71%. The company then announced its inclusion in TOPIX (Oct 7, 2026). For a minority shareholder: there is one share class, no controlling holder and a growing free float. Azbil's 4% stake ties the key supplier into the register. That is useful alignment, but it also means a commercially conflicted shareholder with influence. The unwinding of bank cross-holdings reduces the old "stable shareholder" protection for management.

Part 2 - Governance. The board is small and led by insiders. It is a company with an audit and supervisory committee: 7 directors, of whom 4 are not audit committee members and 3 are outside directors on the audit and supervisory committee. Independent directors therefore make up just under half the board. The President is the representative director, and there is no independent chair. A voluntary nomination and remuneration committee exists, with a majority of independent outside directors. Directors receive restricted stock: after the split, the annual cap on shares granted to eligible directors is adjusted from 80,000 to 320,000 shares. The sources reviewed do not show how strictly that pay is tied to measurable performance. No auditor changes, qualified opinions, insider share pledges or material related-party transactions with management were found. The Azbil relationship is commercial, not a related-party transaction in the filings reviewed. The re-issued Q1 FY3/2027 release in September 2026 was marked "completion of interim review by audit firm". That is a routine step tied to the offering documents, not a restatement.

Part 3 - Capital returns

Dividends. On a split-adjusted basis, DPS was ¥23.0 (FY3/2024), ¥30.5 (FY3/2025) and ¥40.0 (FY3/2026), with ¥56.0 forecast for FY3/2027. That is three consecutive increases, with the payout ratio held at about 30% (31.4%, 30.3% and 30.2%) until the May 2026 switch to a progressive dividend based on a payout ratio of 40% or more or DOE of 7% or more, alongside a higher consolidated ROE target of 15% or more. The FY3/2024 dividend included a ¥20 commemorative dividend.

Buybacks and dilution. Last ~90 days (since Aug 16, 2026): the MoatMap record shows no buybacks. FY3/2024 to date (older than 90 days): searches of the company's disclosures and financial news found no share-repurchase programme. The capital story in this period was two stock splits (2-for-1 in January 2025, 4-for-1 in April 2026, taking issued shares from 16,395,000 to 65,580,000) and small disposals of treasury shares for restricted-stock pay (e.g., Jul 24, 2026). Net of the splits, the share count was essentially flat over three years, with minor dilution from restricted stock. The FY2030 plan allocates ¥20-25bn to shareholder returns (Section 7) but does not commit to buybacks specifically.

Verdict: Returns Capital. The company raised its dividend three years running and has now written a 40%-payout / 7%-DOE progressive floor into policy, but it has not yet used buybacks despite a net-cash balance sheet.

11. Insider Activities

Source: MoatMap disclosure database (Japan large-shareholding reports via EDINET), current as of Nov 13, 2026. Japanese 5%-rule filings carry no per-share price, so values are not available.

Buys - read the signal. No director or officer bought shares in the open market in the window. The only purchases came from Fidelity, a fund manager rebalancing across its funds. That reflects portfolio management, not insider conviction.

Sells - work out the why. Fidelity's June trades were mixed buying and selling across funds, ending with a larger sale on June 22. No reason is disclosed, which is normal for asset-manager filings. The Mizuho filing relates to the bank's exit as part of the company-organised secondary offering, and fits the company's stated review of cross-shareholdings (Section 10). The MoatMap record has no separate line for President Shimada Ryosuke's sale in that same offering. His reduced holding is documented in his own 5%-rule report (Section 10), and the company stated at the time that he would continue to devote himself fully to running the business.

Net assessment. By share count, substantial holders were net sellers over the past year: Fidelity sold roughly 836,000 shares and bought about 270,000, and Mizuho and the President reduced holdings through the October offering. All of the selling has explanations unrelated to the business outlook: fund rebalancing, the unwinding of cross-shareholdings, and an organised offering explicitly designed to widen the free float. No director has bought shares in the open market. Read: neutral, with mild concern that the President chose to sell into the offering, although he remains a significant holder.

12. Scenarios

Bull case. The Academy and the Narashino campus work. Each year's graduate class becomes billable faster, partner-company support keeps field crews intact, and the capacity ceiling rises steadily. Tokyo redevelopment runs to its late-decade end while data centers and reshored factories fill the gap, and every one of those buildings joins the renewal base the company will service for decades. Renewal and maintenance become the dominant business, so a construction slowdown in 2028-2030 hardly shows. Industrial Systems finally reaches meaningful scale through software and electrical works, and a well-chosen acquisition of a regional installer adds engineers rather than overhead. Azbil, needing capacity it cannot hire itself, leans on Nihon Dengi more, not less. In two to three years the company looks less like a contractor and more like a building-services franchise with recurring revenue, a wider shareholder base and a return policy that is paying out more.

Base case. Management delivers roughly what it has guided. The FY2027 plan targets are met or modestly beaten, in line with the pattern in Section 9, as the record backlog converts through successive fourth quarters. Margins drift back from their peak as the unusually profitable projects roll off, just as the President described, while existing-building and maintenance work keeps growing steadily. Data-center and factory demand stays healthy but uneven. Capacity rises slowly because hiring is hard, so growth is capped by headcount rather than demand. Industrial Systems grows but stays a small share. Cross-shareholdings continue to unwind, the dividend rises under the new policy floor, and M&A amounts to small bolt-ons. The business at the end of the period looks like today's, a little larger and a little steadier.

Bear case. Risks 8.1 and 8.4 compound. Partner-company attrition and subcontracting-law costs raise the cost of every project just as the seller's market fades, so the company must accept work on worse terms to keep crews busy. Then 8.3 arrives: the Tokyo pipeline thins after 2028, smaller projects keep slipping on cost inflation, and a pause in data-center spending removes the growth leg. Under 8.5, a large fixed-price job overruns and its handover slips past a March year-end, producing a visibly bad year. Meanwhile 8.6 plays out: the company, flush with cash and committed to deploying a large growth budget, buys an installer or integrator at peak prices that adds headcount but not margin. In the slowest-moving version, 8.2 surfaces as Azbil quietly keeps more flagship projects in-house. The result is a business still solid on its renewal base, but with growth stalled, margins back to contractor norms, and a capital-allocation record that is no longer clean.