1Summary

Takasago Thermal Engineering is the largest listed air-conditioning contractor in Japan. It designs and installs HVAC systems — for offices, stations, department stores and theatres on one side, and for semiconductor fabs, pharmaceutical plants and factory cleanrooms on the other. It does not build the structures. The group also manufactures air-conditioning equipment through a subsidiary, runs a maintenance business, and is spending real money on an attempt to turn green hydrogen into a fourth business line. It is the closest comparator to Sanki Engineering (1961), and the two reported very different quarters.

The year ended 31 March 2026 was strong on both axes, not just margin. Revenue rose 11.1% to ¥423,923m and operating profit rose 47.3% to ¥47,745m, with net income attributable to owners up 35.6% to ¥37,470m. Orders taken rose 10.6% to ¥460,057m. Operating cash flow went from ¥5,885m to ¥29,725m — a fivefold move, which matters more than the earnings line for a contractor.

Then the first quarter of the current year, reported on 5 August 2026, fell across every line: revenue −9.9%, operating profit −16.2%, ordinary profit −11.8%, and net income −28.4%. The company is nonetheless guiding to a full-year increase. That gap is the central thing to understand about this filing, and the reason is visible in the order book rather than in the earnings statement.

Underneath the headline, the mix inverted. At the parent company, general building-equipment orders rose 34.4% to ¥211,819m while industrial-equipment orders fell 13.0% to ¥130,815m. The semiconductor and factory work that drove the last two years is being replaced by metropolitan redevelopment, stations and commercial buildings. Industrial revenue was still rising as that older backlog was worked off (+13.4%), which is why the reported year looks strong and the new quarter does not.

2Business

The group is the parent plus 15 consolidated subsidiaries and 2 equity-method affiliates. It reports two segments and an "other" bucket, and the segments match the legal-entity structure closely.

Equipment Installation (設備工事事業) is effectively the whole company — ¥415,383m of ¥423,923m consolidated revenue after eliminating inter-segment sales. The parent does design and installation of air-conditioning systems; subsidiary TMES handles maintenance and integrated facility management; equity-method affiliate Nippon Setsubi Kogyo does HVAC and plumbing installation. Overseas, wholly or majority-owned arms operate in China (Shanghai and Hong Kong), Singapore, Vietnam, Thailand, Malaysia and Mexico, plus Integrated Cleanroom Technologies in India, which manufactures and installs cleanroom equipment and interior materials. Equipment Manufacturing and Sales is Nippon P-Mac, which designs and makes air-conditioning units. Other is Hu-Cos, an insurance agency.

Two sub-lines are worth separating out because the company discloses them separately and they behave differently from the construction book. Overseas took ¥74,166m of orders (+2.5%) but recognised ¥90,756m of revenue (+26.8%) — overseas revenue exceeded overseas orders, so the overseas backlog drew down during the year. Maintenance took ¥33,975m of orders (+10.5%) on ¥33,299m of revenue (+8.1%), growing steadily and, unlike project work, recurring.

The revenue model is percentage-of-completion contracting. Revenue in any year is largely a function of the backlog taken in prior years and the rate at which it is worked off, not of orders won in that year. This is why the order book, disclosed at parent level, is the forward-looking number and the income statement is the backward-looking one.

The mix shift, in the company's own parent-level disclosure. Takasago splits its construction book into 一般設備 (general building equipment) and 産業設備 (industrial equipment).

Total backlog rose from ¥308,674m to ¥356,812m, up 15.6%. But the composition moved hard: general building equipment is now 71% of the backlog. Public-sector orders more than tripled, from ¥7,233m to ¥24,217m, mostly in general building equipment — municipal offices, prefectural halls, an airport terminal.

Named projects illustrate the switch. The prior year's completed work was led by Rapidus (¥37,447m, 13.7% of parent completed works), the state-backed advanced-logic venture. This year's largest customer was Sony Semiconductor Manufacturing at ¥40,918m, 13.9%. But the new orders named are Tokio Marine's headquarters redevelopment, Shibuya Upper West, Miyagi prefectural hall, Meguro ward office HVAC replacement, and the Shibuya station district development — none of them semiconductor.

Order method. Negotiated (特命) work rose to 65.0% of parent order value from 56.8%, with competitive tender falling to 35.0%. Nearly all of that shift came in general building equipment, where the negotiated share went from 31.3% to 41.9% of the total book. Sanki disclosed a similar move (66.8% to 70.4%). Neither company links this to margin, but both margins rose.

Hydrogen and the fourth domain. This is where Takasago differs from its peers and where the spending is discretionary. The company has been developing hydrogen systems for building equipment for over 20 years. It now sells small water-electrolysis units, and shipped its first megawatt-class unit to Kirin Beer's Hokkaido Chitose plant, with operation scheduled from June 2026 under a ten-year project to convert that plant's fossil-derived city gas to green hydrogen. It operates a microgrid at Atsuta, Ishikari in Hokkaido combining its own electrolyser with solar, batteries and fuel cells, four years into supplying customers. It joined a Tokyo Metropolitan Government project on methylcyclohexane as a hydrogen carrier from April 2026, and signed an agreement with Hida Itsuki Holdings and Inoue Komuten on small-hydro-powered local green hydrogen. R&D was ¥3,931m for the year, against capex of ¥5,030m of which ¥4,887m was BIM-related software development at the parent.

The company also holds a stake in ispace, the lunar-lander company, raised from 587,320 to 6,997,520 shares through an October 2025 third-party allotment and carried at ¥3,043m. The stated rationale is thermal control and water-resource extraction on the lunar surface, and Takasago has been selected for a JAXA Space Exploration Innovation Hub project on lunar heat-pump systems. This is disclosed as a strategic holding, not a portfolio investment.

Cross-shareholdings. At the year-end the company held 34 listed stocks at ¥43,013m and 49 unlisted at ¥2,394m, with an unrealised gain on the marketable portion of ¥31,144m before tax. The listed figure equals 20.0% of consolidated net assets. The stated target is below 15% of net assets by the end of the mid-term plan in FY2026 — a tighter and nearer target than Sanki's sub-20% by March 2028 — and the company notes the ratio rose temporarily this year because share prices went up, not because it bought. It sold ¥1,002m of listed shares and bought ¥3,004m (a rights issue and regular employee-shareholding purchases), so the count is not falling quickly. Largest positions are Kandenko (¥7,726m), Mitsubishi Estate (¥4,407m), Nitto Denko (¥3,064m), ispace (¥3,043m), Hulic (¥2,434m) and Astellas Pharma (¥2,357m). Two are equipment suppliers held explicitly to secure supply — Totech (¥1,201m) and Shinko Kogyo (¥1,043m).

3Results

Full year to 31 March 2026 (consolidated)

Ordinary profit (経常利益) is a Japanese GAAP line between operating and pre-tax profit: it adds recurring non-operating items — interest, dividends, FX, equity-method results — but excludes one-off extraordinary items. For Takasago it runs about ¥2.9bn above operating profit, a wider gap than Sanki's ¥1.3bn, consistent with a larger cross-shareholding portfolio and equity-method affiliates.

Note that net income grew more slowly than operating profit (+35.6% against +47.3%). Ordinary profit grew +44.8%, so the compression is below that line, in tax or in non-controlling interests, not in the operating business.

What management said

The figures below come from the company's own MD&A and supplementary tables in the same securities report (docID S100YDOU) rather than from the machine-verified headline extract, and are presented as the company's disclosure.

Management attributes the profit rise to "smooth construction progress through efforts on an efficient execution structure, in addition to profitability improvement at the order and construction stages" — the same two-part explanation Sanki gave, and like Sanki it does not decompose the margin move further. It describes the market as one where capital investment stayed firm in both manufacturing and non-manufacturing, but where "some customers were taking a careful view on investment timing," and flags materials prices staying high and labour costs rising.

By segment, on the company's basis (which includes inter-segment revenue):

There is no segment going the wrong way here — a contrast with Sanki, where two of four segments declined. The equipment-manufacturing subsidiary grew profit faster than the construction business off a small base.

Balance sheet. Total assets rose ¥46,874m, which the company attributes to higher notes receivable, accounts receivable on completed work, and contract assets — that is, to unbilled and uncollected work, the normal consequence of a rising workload. Liabilities rose ¥16,101m to ¥166,766m, mainly on short-term borrowings. Net assets rose ¥30,773m to ¥215,056m on retained earnings.

Cash flow. Operating cash flow was an inflow of ¥29,725m, up ¥23,839m year on year, which the company attributes mainly to higher pre-tax profit. Investing was an outflow of ¥11,840m, ¥10,435m worse, mainly on purchases of investment securities. Financing was an outflow of ¥16,956m, ¥4,242m worse, mainly dividends. This is the cleanest part of the filing: the profit converted to cash, which it had not in the prior year.

Uses of that cash, as disclosed: capex ¥5,030m, R&D ¥3,931m, and shareholder returns of ¥20,636m — dividends ¥12,634m and buybacks ¥8,002m. Shareholder returns were therefore roughly 55% of net income and more than double the combined capex and R&D. The company also newly established a commercial paper issuance facility during the year and says it is considering committed credit lines and further bond issuance.

First quarter to 30 June 2026 (disclosed 5 August 2026)

Japanese quarterly reporting is cumulative from the start of the fiscal year, so these are three-month figures that become six-month figures in November. They are not annualised.

Company forecast for the full year to March 2027: revenue ¥440,000m, operating profit ¥50,000m, ordinary profit ¥52,000m, net income ¥40,000m. Japanese listed companies publish their own full-year forecast under Tokyo Stock Exchange practice; it is management's number, not an analyst consensus, and revising it is itself a disclosable event.

Set the quarter against that guidance and the arithmetic is demanding. The forecast implies revenue +3.8% and operating profit +4.7% for the year. Q1 delivered −9.9% and −16.2%. Q1 revenue is 19.3% of the full-year forecast, operating profit 17.0%, and net income 16.0%. Construction completions concentrate towards the Japanese fiscal year end, so quarterly progression is never linear — but the guidance requires the remaining nine months to grow, not merely to hold.

The backlog is what supports it. Parent backlog of ¥356,812m at 31 March 2026 against parent completed works of ¥294,497m in the year just ended is 1.2 years of work, up from 1.1 years. The company does not restate backlog in the Q1 release in the securities-report format, which is why the half-year figures in November are the first real test.

4Valuation context

These market figures were retrieved on 13 August 2026 and have not been re-verified for this note. They are stated as arithmetic on disclosed figures and a stated share price, not as a valuation.

At ¥4,270 per share with a market capitalisation of ¥599.8bn (Nikkei):

• Trailing PER: ¥4,270 ÷ ¥285.73 reported EPS for the year to March 2026 = 14.9x
• Forward PER on company guidance: the ¥40,000m net income forecast against the share count implied by Q1 EPS (¥6,408m ÷ ¥48.97 ≈ 130.9m shares, i.e. ex-treasury) gives forecast EPS of about ¥306, or 14.0x. Nikkei quotes 13.9x.
• PBR: ¥599.8bn ÷ ¥215,056m net assets at 31 March 2026 = 2.79x. Nikkei quotes 2.70x, using a more recent book value — Q1 equity is higher, with the equity ratio at 58.9% against 55.0% at year-end.
• Dividend yield: 2.88% on Nikkei's figure.

Peers, all from Nikkei company pages on 13 August 2026, on the same forecast-PER and actual-PBR basis:

The four listed Japanese HVAC and building-services contractors trade within about 1.4 turns of each other on forward PER. The market is not distinguishing between them on earnings multiples, including between Takasago, whose most recent quarter fell, and Sanki, whose most recent quarter roughly doubled at the operating line. The PBR spread is wider and tracks forecast ROE; Taikisha sits apart on both because it combines HVAC with automotive paint-finishing plant.

5What to watch

KPIs, with the level that would matter

1. The general/industrial split of the order book at the interim results. General building equipment orders +34.4%, industrial −13.0%, and the backlog now 71% general. Industrial work is the semiconductor and pharmaceutical plant business — higher-value, lumpier, and the source of the last two years of growth. Whether industrial orders stabilise or keep falling determines what FY2028 revenue looks like, not FY2027.
2. Half-year progress against the ¥440,000m / ¥50,000m guidance. Q1 came in at 19.3% and 17.0% of those numbers with both lines down year on year. Guidance requires growth over the remaining nine months. A first half still tracking below half of guidance would make the full-year number a revision candidate rather than a forecast.
3. Customer concentration. Sony Semiconductor Manufacturing was 13.9% of parent completed works this year; Rapidus was 13.7% last year. Two consecutive years with a single semiconductor customer above 13% is a concentration that the falling industrial order line will resolve one way or the other.
4. Operating cash flow. ¥29,725m this year against ¥5,885m last year. The prior-year figure shows how quickly working capital can absorb the profit in this business. With revenue now falling quarter on quarter, receivables should unwind and support cash — if they do not, that is a collection signal.
5. Policy shareholdings against the 15% target. ¥43,013m of listed holdings is 20.0% of net assets, against a target of below 15% by the end of FY2026 — which is the year now in progress. The company sold only ¥1,002m and bought ¥3,004m this year. Either disposals accelerate materially, or the target is missed; both are readable outcomes in the next annual filing.
6. Hydrogen, at Kirin Chitose. The megawatt-class electrolyser was due to start operation in June 2026 under a ten-year conversion project. This is the first revenue-scale test of the fourth business domain, and the company has committed R&D of ¥3,931m a year against it. Nothing in current earnings depends on it.

Scheduled disclosures

• Half-year (six-month cumulative) results, expected around mid-November 2026 — the first point at which the Q1 decline can be read as a trend or a timing effect.
• Third-quarter cumulative results, expected around early February 2027.
• Full-year results to March 2027, expected around mid-May 2027. This closes the final year of Mid-term Plan 2026, the first phase of the Long-term Vision 2040, and is the reporting date against which the sub-15% policy-shareholding target falls due.
• Guidance revisions are unscheduled but recurrent across this sector.

6Risks the company discloses

Takasago publishes something most filers do not: an explicit rating for each risk on impact (大/中/小), likelihood (高/中/低) and time horizon. That makes management's own ranking legible, so it is worth reading in their order rather than ours.

Rated highest impact (影響度：大). Four items.

Overtime work — likelihood 中, short term. Chronic labour shortage concentrates workload on particular engineers in busy periods. The company states the consequence plainly: reduced productivity and staff attrition shrink execution capacity, and revenue falls as a result. Its response is a shift from site-by-site "construction management" to platform-based "production management" (the T-Base project). Note that this is the same mechanism Sanki describes via the construction-industry overtime cap — available man-hours limit revenue regardless of orders won.

Overseas expansion — likelihood 中, timing unknown. Political instability, war and terrorism, language, legal and tax regimes, trade policy, and unannounced regulatory change across China, Southeast Asia and India. Given that overseas revenue grew 26.8% to ¥90,756m, this is a live rather than theoretical exposure.

Securing people — likelihood 中, medium to long term. Two parts: domestic headcount falling as retirements accelerate, and difficulty recruiting young or specialist staff.

Disasters — likelihood 中, timing unknown. Earthquake, typhoon, tsunami, epidemic. The company notes the two-sided effect: short-term reconstruction demand against medium-term suppression of construction demand overall.

Rated medium impact. Business environment risks come as three linked items: private capital investment being cancelled, delayed or changed in scope; procurement cost and lead-time risk on ducts, piping, insulation, refrigerant and chillers where increases cannot be passed into a signed contract price; and schedule delay from a shortage of engineers and skilled workers at the company and its subcontractors, where the stated consequence is that revenue is simply not recognised. Also rated medium: funding (with the new CP facility as mitigation), accidents and disasters during construction, information security, compliance and litigation, and climate change (medium to long term, transition and physical).

Rated lowest impact but disclosed. Intellectual property. And market risk on assets held — where the company discloses the ¥31,144m pre-tax unrealised gain on its marketable securities and states that a large fall in market value would produce an extraordinary loss. FX exposure is described as limited because the group does little cross-border procurement.

Two absences are worth noting against Sanki, which discloses both. Takasago does not present loss-making contracts as a separately ranked risk, though procurement cost and schedule delay cover parts of the same ground. Nor does it disclose seasonality as a risk in its own right, despite completions clustering at the fiscal year end — a pattern its own Q1 numbers demonstrate.

7Sources

• EDINET annual securities report, docID S100YDOU, filed 2026-06-18 (year ended 31 March 2026). All consolidated headline figures, segment figures, parent-level order and backlog tables, named projects, order-method split, risk ratings, R&D, capex and cross-shareholding disclosure.
• Q1 results release (決算短信) for the year ending March 2027, disclosed 2026-08-05. Q1 figures, full-year company forecast, Q1 EPS and equity ratio.
• Share price ¥4,270, market cap, PBR, dividend yield and peer multiples: Nikkei company pages, retrieved 2026-08-13 and not re-verified for this note — 1969, 1961, 1979, 1980.
• Comparisons with Sanki Engineering are drawn from that company's securities report, docID S100YH3A — see the Sanki Engineering (1961) review.

---

This is a review of reported results. It contains no investment recommendation, no price target and no valuation model. Multiples are stated as arithmetic on disclosed figures and a stated share price.