← EMEGet score and insider alerts

EMCOR Group Deep Dive

IndustrialsGenerated 6 Nov 2026

DEEP DIVE10,000+ word research report

When a new hospital, chip plant or AI data center is built, the concrete and steel are the easy part. What makes the building work is inside the walls.

See EME's live MoatMap Score →Today's Quality, Value and Momentum scores, insider trades, buybacks and financials: the live data behind this report.71/100
12 sections · about 45 min read

Short on time? Start with Key Risks, Walk the Talk or Scenarios.

EMCOR Group, Inc. (NYSE: EME): Deep Dive Research Report

Report date: 6 November 2026. EMCOR's fiscal year ends on 31 December and it reports every quarter. Its latest results, for Q3 FY2026, came out on 29 October 2026, with the earnings call the same day.


1. What the Company Does

When a new hospital, chip plant or AI data center is built, the concrete and steel are the easy part. What makes the building work is inside the walls. Power runs from the utility feed through switchgear and busway to every rack and operating theatre. Chilled water travels through miles of welded pipe to remove the heat the servers produce. Sprinklers, fire alarms, controls, plumbing and ventilation keep people alive and the equipment running. EMCOR installs these systems. Once the building is running, it often stays on to maintain them for decades.

EMCOR is a specialty contractor. A general contractor manages the site and the schedule. EMCOR is the trade firm hired to deliver the electrical and mechanical "guts" of the building. Its 2025 10-K describes about 100 operating subsidiaries and about 44,000 employees, almost all in the United States since the UK business was sold. Most of those subsidiaries are regional contracting firms with local names that EMCOR acquired over three decades and left largely intact. The 2026 ENR Top 600 Specialty Contractors list ranks EMCOR second in the US by specialty-contracting revenue, behind Quanta Services.

Where it came from. EMCOR started as a restructuring. Its predecessor was JWP Inc., which began as Jamaica Water Supply, a water utility serving Queens and Nassau County, New York. Through the late 1980s and early 1990s JWP borrowed heavily to buy more than 100 businesses spanning construction, electronics and computer resale. It filed for Chapter 11 in 1993. EMCOR was formed in December 1994 when JWP emerged from bankruptcy. Frank MacInnis, the CEO of the restructured company, sold off most of the unrelated businesses, including the water utility, and kept the electrical and mechanical contracting core. He then added facilities services so that recurring maintenance revenue would offset the swings of construction.

The milestones that explain today's business are all acquisitions:

  • 2002-2003: 19 operating units bought from Comfort Systems USA, and Consolidated Engineering Services. These built the facilities-services arm that became US Building Services.
  • 2007: FR X Ohmstede, a Texas heat-exchanger and refinery-services business. This is the core of today's US Industrial Services.
  • 2013: RepconStrickland, adding Gulf Coast refinery turnaround capacity.
  • February 2025: Miller Electric of Jacksonville, Florida, for about US$865 million, the largest deal in the company's history. It gave the electrical segment a platform in the Southeast and Texas.
  • 1 December 2025: sale of EMCOR UK to OCS Group, ending the overseas business and leaving a purely US company.

Anthony (Tony) Guzzi has led the company for 15 years. Per the company's proxy disclosures, Guzzi joined EMCOR as President and Chief Operating Officer in October 2004 and became CEO on 3 January 2011. Before that, from 1997 to October 2004, Guzzi held senior roles at United Technologies and its Carrier subsidiary, and from August 2001 ran Carrier's North American Distribution and Aftermarket division. Guzzi joined the board in December 2009 and became Chairman on 1 June 2018.

The value proposition. An owner building a US$10 billion data-center campus needs three things from its electrical and mechanical contractor.

  1. Enough supervised, skilled craft labour, available on the owner's schedule, often in a rural county with no local workforce.
  2. Confidence that the systems will be commissioned on time. A data hall that cannot be energised or cooled earns nothing.
  3. A safety record that keeps the site open. EMCOR's 2025 total recordable incident rate was just under 1.0, about 60% below the latest US industry average of 2.4, and below average for the 17th consecutive year (2025 10-K).

EMCOR's offer is that it can provide these at the same standard in many markets at once, and then service the systems afterwards.

Why it is hard to replicate. The constraint is people, not technology. More precisely, it is the foremen, general foremen and project managers who turn a crew of electricians or pipefitters into a productive team. Management named supervisory talent as the binding constraint on growth (Q1 2026 call, 29 April 2026). Building a bench of supervisors takes years, and EMCOR has assembled one through acquisitions and union apprenticeship systems. The prefabrication and design capability that multiplies that labour is covered in Section 3.

A concrete example. A hyperscaler is building a multi-building AI training campus in central Texas. The owner buys the big equipment directly: switchgear, generators, chillers and transformers. EMCOR's electrical subsidiary in the region, which in Austin is now Schmidt Electric (acquired 2026), lays out and installs the medium- and low-voltage distribution, cable tray and busway, grounding, lighting, fire alarm and low-voltage data systems. A mechanical subsidiary fabricates pipe racks and piping modules in a shop, ships them to site and installs the chilled-water plant and the liquid-cooling distribution that AI racks increasingly require. A fire-protection unit designs and installs sprinklers. Each building is a phase, and EMCOR books only the phases it has under contract. Once the campus is energised, the building-services arm can bid for long-term HVAC and controls maintenance.


2. Business Segments

EMCOR reports four US segments. It reported a fifth, UK Building Services, until it sold that business on 1 December 2025. The FY2025 mix below comes from full-year segment revenue in the Q4 2025 release. The year-to-date 2026 mix comes from the Q3 2026 release. The mix is shifting quickly toward the two construction segments, so both are shown.

SegmentFY2025 share of revenue9M 2026 shareRole in the group
US Mechanical Construction & Facilities Services~42%~45%Largest segment; the data-center cooling engine
US Electrical Construction & Facilities Services~30%~32%Fastest-growing; platform for M&A
US Building Services~18%~16%Recurring service base
US Industrial Services~7.5%~7%Cyclical oil-and-gas services
UK Building Services (sold Dec 2025)~3%-Exited

US Mechanical Construction & Facilities Services (~42% of FY2025 revenue)

This segment installs and fabricates the systems that move air, water and heat:

  • HVAC
  • process and chilled-water piping
  • plumbing
  • fire protection and sprinklers
  • clean-room ventilation
  • geothermal systems
  • building controls
  • sheet-metal ductwork
  • millwright work

Its customers are general contractors and owners in data centers, high-tech manufacturing (semiconductor, pharma, biotech), healthcare, universities and other institutions, warehousing, and water and wastewater treatment.

Year-to-date 2026, network and communications work (EMCOR's label for data centers) was about 35% of the segment's revenue. Commercial was 12%, manufacturing and industrial 11%, high-tech manufacturing 9%, institutional 8%, service work 8% and healthcare 7%.

Core capability. Large-diameter pipe welding and the off-site fabrication of piping and ductwork at scale. Mechanical work on an AI data center is the most fabrication-heavy trade on the site. On the Q2 2026 call (30 July 2026), Guzzi said AI facilities carry a 1.5-2x multiplier on mechanical content compared with a traditional cloud data center, because liquid cooling and denser heat loads require far more piping.

Why it is separate. Different craft unions (pipefitters, plumbers, sheet-metal workers), different fabrication assets, and a different acquisition lineage from the electrical firms.

Fit. The largest segment and the main channel for AI-driven growth. Its revenue nearly doubled from data centers in 2026. Management is accepting more construction-manager and cost-plus roles here. Those contracts carry lower percentage markups, a trade-off examined in Section 8.

Main rivals: Comfort Systems USA, Limbach, APi Group (fire protection) and many private regional mechanical firms.

US Electrical Construction & Facilities Services (~30% of FY2025 revenue)

This segment installs:

  • power distribution, from medium voltage down to branch circuits
  • premises wiring and lighting
  • low-voltage, voice and data systems
  • fire alarm and security systems
  • sustainable-energy installations
  • some power transmission and distribution work

Year-to-date 2026, data centers were 56% of the segment's revenue. Commercial, manufacturing and industrial, healthcare and institutional were each about 7%.

Core capability. Coverage. Guzzi said EMCOR can serve data-center customers electrically in about 17 markets, which few others can (Q4 2025 call, 26 February 2026). The electrical segment is also where EMCOR buys most often. Miller Electric (2025) and the 2026 cluster of B&B Electric (Wisconsin), Sidney Electric (Ohio), Giles (Daytona Beach), Schmidt Electric (Austin) and Connelly Electric (Chicago) all landed here.

Why it is separate. A different craft (IBEW electricians), different customer relationships and different labour economics from mechanical.

Fit. The growth platform. Management uses electrical acquisitions to enter a metro area, then expands into data-center work there. Electrical revenue rose about 52% in 2025, helped by Miller.

Main rivals: Quanta Services (after buying Cupertino Electric), IES Holdings, MYR Group, Rosendin Electric (private) and regional electrical contractors.

US Building Services (~18% of FY2025 revenue)

The 2025 10-K describes three businesses.

  1. Mechanical services. HVAC service agreements, repairs, retrofits, controls upgrades, indoor-air-quality and energy-efficiency work for existing buildings.
  2. Commercial site-based services. On-site facilities management, maintenance, floor care and landscaping.
  3. Government services. Contracts with federal agencies, including the National Archives and the FDIC, with base periods and option years.

Core capability. Dense local branch networks and technicians who know the installed equipment. Mechanical services competes with OEM service arms (Carrier, Trane) and with roll-ups such as Service Logic.

Why it is separate. Recurring, contract-based revenue with low capital needs and different customers: building owners and property managers, not general contractors.

Fit. The stabiliser. Management restructured the site-based business in 2024-25 and calls it leaner, and it wants more mechanical-services acquisitions. In Q3 2026 EMCOR wrote down the intangible assets of an energy-efficiency contracting business in this segment, which management described as under 1% of segment revenue, after disruption in the performance-contracting market (Q3 2026 call, 29 October 2026).

Main rivals: ABM Industries, CBRE, JLL, Cushman & Wakefield, Carrier and Trane service arms, and Service Logic.

US Industrial Services (~7.5% of FY2025 revenue)

This segment has two parts.

  • Field services. Refinery and petrochemical turnarounds (planned shutdowns for inspection and repair), specialty welding and critical-equipment overhaul.
  • Shop services. Designs and manufactures shell-and-tube heat exchangers and repairs them in its own shops (the Ohmstede legacy).

It also does some renewable-energy work; a large solar project lifted Q2 2026.

Core capability. Gulf Coast turnaround crews and engineered heat-exchanger manufacturing.

Why it is separate. Entirely tied to oil, gas and petrochemical maintenance cycles. Its economics depend on refinery margins and turnaround timing, not on building construction.

Fit. A cyclical option. It is the most volatile segment: an operating loss in Q2 2025, then profit nearly doubling on a better shop mix in Q3 2025. Management describes it as a business to run well, not one to grow aggressively.

Main rivals: Team Inc., Turner Industries, Universal Plant Services, Dunn Heat Exchangers and other firms named in the 10-K.

Geographic mix and the UK exit

In FY2025, EMCOR earned about 97% of revenue in the US and 3% in the UK (2025 10-K). The UK business, a facilities-services operation inherited from the JWP era, was sold to OCS Group. Management put the price at about US$255 million when announcing the sale (Q3 2025 call, 30 October 2025). The deal closed on 1 December 2025. From 2026 EMCOR is a purely US company.

Within the US, its data-center footprint is concentrated in Northern Virginia, the largest single concentration of data-center backlog, and Texas, the second largest (Q2 2026 call). Growing data-center markets include Georgia, Ohio, Iowa, Oklahoma and Pennsylvania (Q3 2026 call). Water and wastewater growth is concentrated in Florida (Q1 and Q3 2026 calls). The Miller acquisition made the Southeast a core region for the first time.


3. Products and Operations

What EMCOR actually installs

Seen as an engineer would see it, EMCOR's output falls into six system families.

  1. Power distribution (electrical). From the owner's switchgear and transformers down to busway, panelboards, branch circuits and rack-level power. In data centers this includes redundant distribution, typically two independent paths to every rack, and connections to generators and UPS systems. The owner buys the major equipment itself. Management noted EMCOR faces no equipment bottlenecks for that reason (Q1 2026 call).
  2. Low-voltage and life-safety systems (electrical). Structured cabling, fire alarm, security and access control, and building automation wiring.
  3. HVAC and process cooling (mechanical). Chilled-water plants, cooling towers, air handlers, computer-room air handlers and, increasingly, liquid-cooling distribution (coolant distribution units and secondary piping loops) for AI racks.
  4. Process and specialty piping (mechanical). High-purity piping for semiconductor and pharma plants, and process piping for food, industrial and water treatment plants.
  5. Fire protection (mechanical). Sprinkler design, fabrication and installation, followed by recurring inspection and maintenance. Guzzi described EMCOR as one of the few contractors able to deliver fire and life safety nationally (Q4 2025 call).
  6. Heat exchangers (industrial). Engineered shell-and-tube units built in EMCOR's own shops, plus repair and re-tubing during refinery turnarounds.

Why AI changes the content per building

Data-center projects are getting bigger. Guzzi described typical data-center projects growing from about 20 MW historically to 40-75 MW, with AI facilities at 100-200+ MW and full campuses beyond that (Q2 2026 call, 30 July 2026). Content per megawatt is rising too. AI facilities carry roughly 1.5-2x the mechanical content and roughly 1.5x the electrical content of a traditional cloud facility (Q2 2026 call).

Management estimates EMCOR's data-center work is currently about 65-70% cloud and storage and about 30% AI training, and expects the mix to move toward 50/50 over time (Q3 2026 call). The mechanical segment has the highest AI share because the cooling load is where AI facilities differ most.

How the work is delivered

Bid / negotiate -> Design assist (VDC/BIM) -> Off-site fabrication -> Field installation -> Commissioning -> Service contract

Design assist and virtual design and construction (VDC). EMCOR usually is not the engineer of record. It picks up the design at about 60-70% completion and finishes it for "constructability": a 3D model detailed down to every hanger and pipe joint, clash-checked against the other trades, so that components can be fabricated before anyone reaches the site (Q2 2025 call, 31 July 2025).

Prefabrication. About 95% of EMCOR's fabrication output goes to its own jobs and about 5% to third parties (Q2 2025 call). Shops produce pipe spools and racks, multi-trade modules, ductwork, electrical assemblies and sprinkler mains. Prefab matters for three reasons:

  • shop work is safer and more productive than field work
  • it cuts the number of craft workers needed on remote rural sites, which is how EMCOR can staff data centers in counties with no local labour (Q2 2026 call)
  • it raises the content EMCOR captures on mechanically heavy AI jobs

Capital spending has been growing about twice as fast as revenue for several years, mostly on prefab. Planned 2026 capex is US$115-125 million (CFO Jason Nalbandian, Q1 2026 call).

On the Q3 2026 call (29 October 2026), Guzzi said EMCOR is adding fabrication-only capacity in well-placed locations to build piping racks, HVAC modules and similar assemblies for hyperscale customers. Guzzi drew a clear line on what this is:

"we are not building buildings and then putting them on tractor trailers and moving them out" - Tony Guzzi, Q3 2026 call, 29 October 2026

Field installation and labour. About 62% of EMCOR's roughly 44,000 employees are union members under about 450 collective-bargaining agreements (2025 10-K). The union hall is effectively EMCOR's flexible workforce. Electricians, pipefitters and sheet-metal workers come off the hall for a project and go back when it ends, and union apprenticeships train them to a common standard. Craft workers also travel between regions. Management said traditional sites run four- to five-day weeks while data-center sites offer weekend overtime, which helps attract travelling labour (Q3 2026 call). Guzzi has left room for non-union operations in very rural markets but called the union system more sophisticated and predictable (Q2 2026 call).

Market entry. EMCOR adds roughly one or two new geographic data-center markets a year. It moves supervisors in, hires locally and accepts lower productivity at first. Guzzi called this startup cost "a one or two quarter headwind," treated like routine R&D for expansion (Q3 2025 call). In 2025 the electrical segment absorbed exactly that cost on projects in newly entered markets.

Service after installation. Building Services sells maintenance agreements, retrofits, controls upgrades and repairs on installed systems. Its 2026 growth has come from HVAC retrofits, controls and indoor-air-quality work, plus new facilities-maintenance contracts at the site-based business (Q2 and Q3 2026 calls).

Industrial delivery. Field crews mobilise for refinery turnarounds, typically in spring and autumn, working around-the-clock shifts during short planned shutdowns. Shops along the Gulf Coast manufacture and repair heat exchangers. Turnaround timing is set by refiners and moves between quarters: some 2025 turnarounds slipped into late 2025 and 2026 (Q3 2025 call).

Recent capacity and footprint events

  • Acquisitions closed in 2025. Miller Electric (February) and John W. Danforth Company (Q4 2025, a mechanical contractor management said would add US$350-400 million of annual revenue at signing).
  • Acquisitions closed in 2026. Nine companies for US$809.9 million upfront in the first nine months, including the five electrical firms named in Section 2. Management put those five at about US$750 million upfront plus up to US$90 million in earn-outs (Q2 2026 call). Two more companies closed in October 2026 for US$72 million (Q3 2026 10-Q).
  • Fabrication. The fabrication-only capacity expansion described above, announced on the Q3 2026 call.

4. Customers

Who buys

No single customer accounts for more than 10% of revenue (2025 10-K). The spread across end markets is wide, but data centers now dominate the construction book. Year-to-date 2026, network and communications was about 44% of combined electrical and mechanical construction revenue. Manufacturing and industrial and commercial were about 10% each, institutional and healthcare about 7-8% each, high-tech manufacturing about 6%, and service work, short-duration projects, water, hospitality and transportation made up the rest (Q3 2026 10-Q disaggregation). The main customer types follow.

Hyperscalers and the developers who build for them

Who decides. Hyperscalers' data-center construction groups, and developers building for them on long leases. EMCOR is usually hired through a general contractor but is often selected or influenced by the owner. Guzzi said three of the four largest hyperscalers shared proprietary multi-year capital plans with EMCOR (Q3 2025 call). That points to a relationship where the owner is planning labour years ahead. EMCOR says it works for "the big hyperscalers and those serving the bigger hyperscalers" and deliberately avoids speculative developers (Q3 2026 call).

Criteria. Schedule certainty, available labour, safety and the ability to repeat a design across multiple sites and markets. Price matters, but differently than on a normal building, because the cost of a late energisation is far larger than the cost of the contractor. Guzzi's description of these customers (Q2 2025 call) is a useful reality check: they are the most sophisticated and demanding customers in the market and "not paying one cent more than needed."

Sales cycle. Multi-year relationships, with individual phases awarded in sequence. Larger awards increasingly come as GMP (guaranteed maximum price) or cost-plus contracts with an evolving scope. EMCOR books only the contracted phase, even when follow-on phases are highly likely (Q3 2025 call).

Healthcare systems, universities and institutions

Hospital systems modernising facilities and universities upgrading campuses are buyers with long capital plans and complex, regulated systems: medical gases, infection-control ventilation and critical power. Facilities and capital-projects departments make the decisions, often through construction managers. Sales cycles run from 12 to 36 months. Institutional revenue in the mechanical segment rose 77% in Q2 2026. Healthcare and institutional backlog both grew strongly through 2026 (Q2 and Q3 2026 calls).

Water and wastewater utilities

Municipal and regional utilities, concentrated in Florida, buy treatment-plant mechanical and process work. In this sector EMCOR sometimes acts as prime contractor, which management flagged as a lower-margin mix item when setting 2026 guidance (Q4 2025 call). Procurement is public: competitive bids or qualifications-based selection.

Manufacturers, including semiconductors

High-tech manufacturing (semiconductor fabs, pharma, biotech, EV batteries) buys clean-room ventilation, high-purity process piping and fire and life safety. Management cited fire-protection and mechanical opportunities at Arizona semiconductor sites (Q2 2026 call). EV-related work has declined and AI-related semiconductor work has partly replaced it (Q3 2026 call).

Building owners and government agencies (Building Services)

Property managers, corporate real-estate departments and federal agencies buy maintenance agreements, retrofits and facility management. Decisions sit with facilities directors. Contracts are often multi-year. Federal contracts have base periods plus option years (2025 10-K).

Refiners and petrochemical operators (Industrial Services)

Plant maintenance and turnaround planners at refiners and chemical plants buy turnaround labour and heat-exchanger work. Turnarounds are planned 12-24 months ahead, but refiners move them for operating reasons. Guzzi warned that disruption tied to the Middle East would make refiners keep units running instead of shutting them for maintenance (Q2 2026 call).

Switching costs

Switching costs differ sharply by business.

  • Construction. Weak to moderate on any single project. Each job is awarded separately, and a general contractor can pick another trade partner. Higher switching costs come from two sources. First, a contractor that has completed several phases of a campus knows the design, the owner's standards and the local crews, so replacing it adds schedule risk. Second, prequalification: hyperscalers and hospital systems keep approved-contractor lists, and getting onto one requires a safety record, bonding capacity and references.
  • Building Services. Moderate. The service technician who knows a building's chillers and controls is hard to replace without disruption, but service agreements are re-bid on a cycle.
  • Industrial Services. Moderate. Shop-built heat exchangers are engineered to fit the plant, and repair work tends to stay with the shop that knows the unit.

Contract structure and predictability

In 2025, projects worth US$10 million or more made up about 58% of electrical and mechanical construction revenue, typically ranging up to and sometimes beyond US$200 million. Smaller projects made up 42% and usually finish within a year (2025 10-K).

Fixed-price contracts are still the default on most construction work. GMP and cost-plus arrangements have grown on data-center mechanical work, where scope keeps changing (Q1 2026 call). Remaining performance obligations count only contracted, non-cancellable work and approved change orders (Guzzi, Q3 2026 call). Roughly 78-82% of RPO converts to revenue within 12 months (Q4 2025 and Q1 2026 calls).

The result is very good 12-month visibility and little contracted visibility beyond that. Building Services adds a layer of recurring revenue on top, though it is a shrinking share of the total.

Concentration

There is no customer concentration in the reporting sense, but there is end-market concentration. Data centers made up nearly half of total backlog at the end of Q3 2026, and a small number of hyperscalers sit behind most of that demand. That reflects how far EMCOR has been allowed into its customers' plans. It is also the largest single risk to the business (Section 8.1).


5. Competitive Landscape and Moat

Industry structure

US electrical and mechanical contracting is very fragmented. The ENR 2026 Top 600 Specialty Contractors together reported US$275.1 billion of revenue, up 14.7%, which still captures only part of a market served by tens of thousands of local firms. A few large companies operate nationally:

  • Quanta Services, the No. 1 specialty contractor (ENR 2026)
  • EMCOR, No. 2
  • Comfort Systems USA, No. 4
  • a few private giants such as Rosendin Electric

Beneath them are hundreds of strong regional firms, many of which are acquisition targets for EMCOR, Comfort Systems, Quanta, IES and APi.

Below the national tier, competition is local. A data-center job in Northern Virginia is fought out among the firms with crews in Northern Virginia. EMCOR's scale advantage therefore comes not from being big nationally but from holding a top-tier position in many local markets at the same time, and from being able to move supervisors and processes between them.

CompetitorCountryListingApprox Market CapProduct OverlapRelative Strength
Quanta ServicesUSANYSE: PWR~US$77bn (Nov 2026)Electrical; data-center premises electrical via Cupertino Electric; utility T&DFar larger in utility transmission and grid; overlaps EMCOR inside the data-center fence
Comfort Systems USAUSANYSE: FIX~US$34bn (Nov 2026)Mechanical (HVAC, plumbing, modular), electrical, serviceClosest like-for-like rival; strong in modular mechanical and data centers
APi GroupUSANYSE: APG~US$17bn (Nov 2026)Fire protection and life-safety installation and inspectionLeader in recurring fire-safety inspection; overlaps EMCOR's fire protection
IES HoldingsUSANasdaq: IESC~US$6bn (Nov 2026)Electrical, data-center communications, modular power enclosuresStrong in data-center electrical and prefab power modules
MYR GroupUSANasdaq: MYRG~US$3.8bn (Nov 2026)Commercial and industrial electrical; T&DSmaller, more T&D-weighted
ABM IndustriesUSANYSE: ABM~US$3.1bn (Nov 2026)Facilities services, engineering, HVAC serviceLarger in janitorial and facility services; weaker in technical trades
Limbach HoldingsUSANasdaq: LMB~US$0.6bn (Nov 2026)Mechanical construction and owner-direct serviceNiche, focused on owner-direct work
Rosendin ElectricUSAPrivate-Electrical, data centers, renewablesLarge employee-owned electrical contractor; ENR West 2026 Specialty Contractor of the Year
Service LogicUSAPrivate-Mechanical services roll-upPrivate-equity-backed HVAC service competitor

Market caps from CompaniesMarketCap, 5 November 2026; peer-size reference only.

Where EMCOR wins and loses

  • Against Comfort Systems USA. Comfort is the most direct comparison: a US roll-up of mechanical and electrical subsidiaries that has moved heavily into data centers and modular construction. EMCOR wins on electrical breadth (17 data-center electrical markets) and on fire and life safety nationally. Comfort has historically been more mechanical and more modular. The two firms pursue the same hyperscaler work and the same acquisition targets, so competition between them shows up in deal pricing as much as in project bids. Comfort sold EMCOR the units that formed part of today's Building Services segment in 2002.
  • Against Quanta. Quanta wins anything tied to the utility grid: transmission lines, substations, interconnection. Inside the fence it competes with EMCOR in premises electrical, mainly through Cupertino Electric. EMCOR has the mechanical, fire and service breadth that Quanta lacks. A hyperscaler can split the work, using Quanta for the grid connection and EMCOR for the building.
  • Against IES and MYR Group. Both are smaller and more focused. IES has a strong data-center electrical and modular-power business. EMCOR's advantage is multi-trade breadth and geographic reach. For these smaller firms, competition with EMCOR is mostly local.
  • Against private regional firms. This is where EMCOR wins most and loses most. A dominant private contractor in a single metro area can beat EMCOR there on relationships and price. EMCOR's answer is often to buy that firm, as with Schmidt in Austin and Connelly in Chicago.
  • In Building Services, EMCOR competes against facilities-management giants (CBRE, JLL, Cushman & Wakefield, ABM) at the site-based business and against OEM service arms (Carrier, Trane) and private roll-ups at the mechanical-services business. EMCOR has no structural advantage at the site-based business and a moderate one in mechanical services, where its technicians' technical depth matters.

Barriers to entry

Barriers are low for a small local contractor and high for a national one serving hyperscalers. A new entrant at that tier needs four things.

  1. Bonding capacity and balance sheet. Sureties will not back large GMP projects without a strong balance sheet.
  2. Supervisory depth. The foremen and project managers EMCOR identifies as its constraint (Section 1). There is no quick way to create them.
  3. Safety record and prequalification. A recordable-incident rate well below the industry average is a condition for getting on hyperscaler and hospital bid lists.
  4. Union and labour access in each market.

None of these depends on intellectual property, and all of them can be bought. That is why the competitive threat comes from well-funded roll-ups and private-equity platforms buying regional firms, not from start-ups.

Market share

Even in its strongest segments, EMCOR's national share is in the low single digits. The relevant share is local: in the markets where it holds data-center electrical or mechanical work, it is one of a few contractors able to do the largest jobs. Share is fragmented because construction labour is local and union jurisdictions are regional. Consolidation happens one acquisition at a time.

Structural shifts

  • Consolidation is accelerating. EMCOR bought at least 11 companies in 2026 alone. Comfort Systems, IES, APi and Quanta are all acquiring regional firms. The pool of high-quality independent contractors is shrinking, and prices for those with data-center exposure are rising. Guzzi said EMCOR will not pay for "one market company" valuations (Q1 2026 call).
  • Prefabrication and modularisation shift value from field labour to shop capacity and engineering. That favours large contractors with capital to spend, and invites manufacturers of modular power and cooling units into the trade contractor's space.
  • Contract forms are moving toward GMP and cost-plus on hyperscale work. This lowers fixed-price risk but compresses percentage markups.
  • Owners are going direct. Some hyperscalers now contract trade partners directly and plan labour across multiple years. That rewards the few contractors able to cover many markets.

Moat: Narrow - scale / cost advantage (multi-market supervised labour and prefabrication) and qualification barriers (safety record, bonding, hyperscaler prequalification)

Section 4 shows that hyperscalers share multi-year plans with a few contractors and avoid replacing a proven partner on an active campus. Section 5 shows that only a handful of firms can staff electrical, mechanical and fire work across many markets at once. Those advantages are real, but they have not been tested in a well-funded attack or a severe construction downturn in their current data-center-heavy form. Every element of the advantage can also be bought through acquisitions. The thing most likely to erode it is competitors and private-equity roll-ups building comparable multi-market supervisory benches, combined with hyperscalers using their purchasing power to push GMP terms toward thin markups.


6. Industry

What drives demand

Demand for electrical and mechanical contracting follows non-residential building construction, plus a maintenance and retrofit layer on the existing building stock. Several forces drive it.

  1. Digital infrastructure. Cloud and AI data centers. This is the dominant force in the current cycle.
  2. Industrial policy and reshoring. Semiconductor fabs, pharma plants and battery plants, supported in recent years by federal incentives.
  3. Healthcare and institutional capital cycles. Ageing hospitals and campuses, and public-sector funding.
  4. Water infrastructure. Population growth in the Sun Belt and federal infrastructure funding.
  5. Building performance. Energy codes, decarbonisation, indoor air quality and controls retrofits on existing buildings.

Size and trajectory

According to the AIA Consensus Construction Forecast (July 2026), US non-residential building spending fell about 2% in 2025 to roughly US$846 billion. The panel cut its 2026 forecast from +1.0% in January to -0.3% in July, and expects +3.0% in 2027. Through the first five months of 2026, the sector was down about 7% year on year.

The AIA calls the market "K-shaped". Data-center spending is forecast to rise about 33% in 2026 and to reach roughly 8% of all non-residential building spending by 2027. Manufacturing construction is forecast down 11.6% in 2026 as the reshoring wave of 2022-24 fades. In Census data, data centers sit inside the "office" category, so office spending growth reflects data centers propping up a shrinking conventional office market.

The ENR Top 600 Specialty Contractors grew revenue 14.7% in their most recent reporting year, a much faster rate than the overall construction market, as AI infrastructure absorbed skilled labour.

The industry is therefore two markets. One is a data-center market growing at a double-digit rate. The other is everything else, which is broadly flat to declining in real terms, with healthcare, institutional and water as the exceptions. Management claims EMCOR has historically grown organically about 600 basis points faster than non-residential construction, and its construction segments 700-900 basis points faster (Q3 2026 call).

Position in the supply chain

Owner / hyperscaler -> Developer -> General contractor / construction manager -> Specialty trade contractors (EMCOR) -> Craft labour (union halls) ^ Equipment OEMs (switchgear, chillers, generators, UPS) - procured by owner -------+

EMCOR sits in the specialty-trade layer. It turns owner-procured equipment and engineer-of-record designs into working systems. Owners buy most of the high-value equipment directly, so EMCOR's revenue is mostly labour, fabrication, materials (pipe, wire, conduit, duct) and management.

Regulation

  • Licensing and codes. State and municipal electrical and mechanical licences, the National Electrical Code and fire codes, inspected by local authorities.
  • Labour. Collective bargaining, prevailing-wage rules on public projects, and multi-employer pension plans such as the National Electrical Benefit Fund and the National Automatic Sprinkler Industry Pension Fund (2025 10-K).
  • Data-center siting and power. Local zoning, tax treatment and utility interconnection increasingly decide where data centers are built. Loudoun County, Virginia paused approvals in 2026. Guzzi argued that local tax revenue makes moratoriums hard to sustain and that most projects due in the next two to three years are not affected (Q2 and Q3 2026 calls).
  • Tariffs and trade. Tariffs on steel, copper and equipment raise materials costs and create pricing uncertainty, which management cited when setting guidance (Q2 2025 call).

Cyclicality

Non-residential construction is late-cycle. It follows business investment and credit conditions with a lag of 12-24 months, because projects are funded and designed well before they are built. Specialty contractors feel a downturn when bidding becomes more competitive and margins compress before revenue falls, since backlog keeps work flowing while new awards get cheaper.

Maintenance and service work is much steadier: buildings must be maintained in a recession, though discretionary retrofits get deferred. Refinery turnarounds follow refiner economics and run on their own cycle.

Tailwinds and headwinds

Tailwinds

  • AI data-center buildout with more electrical and mechanical content per MW
  • Water infrastructure in the Sun Belt
  • Healthcare and campus modernisation
  • Energy-efficiency and controls retrofits
  • Industry consolidation, which rewards scaled acquirers
  • Skilled-labour scarcity, which raises the value of contractors who have labour

Headwinds

  • Commercial office and general non-residential weakness
  • Fading manufacturing construction after the reshoring peak
  • Power availability limiting where data centers can go
  • Local opposition to data centers
  • Tariff-driven materials inflation
  • The skilled-labour shortage itself, which caps how fast the industry can deliver

7. Growth Triggers

Ranked by potential impact on the business.

  • Conversion of a doubled data-center backlog, with a shift toward AI facilities that carry more content per MW. Network and communications backlog more than doubled year on year to a record by the end of Q3 2026. Management expects the mix to move from roughly 65-70% cloud and 30% AI toward 50/50, with AI jobs carrying 1.5-2x the mechanical content and about 1.5x the electrical content (Q2 2026 call, 30 July 2026; Q3 2026 call, 29 October 2026). This repeats across every call in the window. Management said demand had shown "none" change in profile (Q2 2026) and that it saw "no sign of slowing" (Q1 2026 call, 29 April 2026). Campuses of 100-200+ MW are expected to arrive as multi-phase awards.
  • New data-center geographies. EMCOR serves data centers in about 17 electrical markets and 7 mechanical markets (Q4 2025 call, 26 February 2026) and expects to add one or two markets a year (Q3 2025 call, 30 October 2025). Management named Texas, Georgia, Ohio, Iowa, Oklahoma and Pennsylvania as growth markets alongside Virginia (Q3 2026 call). Rural sites served through prefab are a stated priority (Q2 2026 call).
  • 2026 electrical acquisitions moving into data-center work. Schmidt (Austin), Connelly (Chicago), B&B (Wisconsin), Sidney (Ohio) and Giles (Daytona Beach) were bought partly to move from warehousing, industrial and space work into data centers (Q2 2026 call). Management expects acquired backlog to keep contributing through 2027 (Q3 2026 call). The acquisition pipeline was described as "broader and more diverse" than a year earlier, with targets from US$2 million to more than US$500 million in electrical, mechanical services, fire protection and building controls (Q4 2025 call; repeated Q1 2026).
  • Fabrication-only capacity for hyperscale components. New shop capacity in well-placed locations to build piping racks, HVAC modules and similar assemblies at a larger scale for hyperscale customers (Q3 2026 call). This adds to continuing prefab capex (Q1 2026 call).
  • Healthcare, institutional and water backlog coming due. Healthcare, institutional and water and wastewater backlog each grew roughly 25-40% year on year at Q3 2026. Florida water treatment and university projects were highlighted (Q4 2025 call; Q3 2026 call). Repeated across the four most recent calls.
  • Manufacturing and high-tech recovery led by AI-related semiconductor work. Manufacturing and industrial backlog grew sharply quarter on quarter in Q3 2026. AI-related semiconductor work is starting to replace EV battery projects, with Arizona fab fire-protection and mechanical work named as opportunities (Q2 2026 call; Q3 2026 call).
  • Building Services: retrofits, controls and new facilities-maintenance contracts. HVAC retrofits, controls upgrades and indoor-air-quality work in mechanical services, and new site-based contracts after the restructuring (Q2 2026 call; Q3 2026 call). Building controls and mechanical-services acquisitions are a stated M&A focus (Q4 2025 call).
  • Industrial Services: LNG maintenance and petrochemical work. Management pointed to LNG maintenance and petrochemical opportunities, with turnaround timing and Middle East disruption as the swing factor (Q3 2026 call; see Section 9 for the Q2 2026 warning).

8. Key Risks

8.1 A pause in data-center capex hitting nearly half the backlog

Mechanism. Data centers are about 44% of year-to-date 2026 construction revenue and nearly half of backlog (Section 4). A small number of hyperscalers and the developers serving them account for most of that demand. Hyperscalers set capex annually and can cut quickly when AI monetisation disappoints, capital costs rise or chip and power supply shifts. Remaining performance obligations include only contracted phases, so a pause would appear first as follow-on phases not being awarded. Revenue would hold up for two to four quarters while new bookings dried up. Labour and supervisors would then be stranded in the markets EMCOR entered for data-center work, especially the newer rural sites.

Management's view:

"We see no sign of slowing demand in this vertical, where customer investments in AI infrastructure, cloud infrastructure and overall digital transformation are driving unprecedented levels of activity." - Tony Guzzi, Q1 2026 call, 29 April 2026

Management has also said data-center share of backlog may fluctuate and should not be read as a deliberate cap (Q4 2025 call). In other words, there is no stated limit on concentration.

Calibration. Moderate probability over three years, high impact. The AI capex cycle has never been through a downturn. EMCOR's backlog quality is high because it avoids speculative developers, but a slowdown among high-quality customers still means fewer new phases.

8.2 Contract mix and new-market execution diluting returns

Mechanism. Two separate pressures.

  • Contract form. More data-center mechanical work is coming as construction-manager, GMP and cost-plus contracts. These protect EMCOR against cost overruns but carry lower percentage markups. Mechanical margins fell year on year in both Q2 and Q3 2026 for this reason (Q2 and Q3 2026 calls).
  • Startup losses. Each new market costs one or two quarters of low productivity while local crews are built. In Q3 2025 the electrical segment absorbed lower profitability on projects in newly entered markets. With one or two new markets a year, plus larger and more complex AI jobs on fixed-price terms where they still apply, a single bad project (a schedule slip, a craft shortage, a design change disputed after the fact) could wipe out the profit on several good ones.

Management is open about what it is optimising for:

"We're much more focused on growing margin dollars, which is what you actually spend and invest for the long term." - Tony Guzzi, Q1 2026 call, 29 April 2026

Calibration. High probability of moderate dilution. Low probability that a single project loss is large enough to matter at group level, given how many projects EMCOR runs.

8.3 The supervisory and craft-labour bottleneck, and the union-pension tail

Mechanism. Growth is limited by foremen and project managers (Section 1). If data-center demand outruns EMCOR's supervisory bench, there are three bad options: turn work down, overstretch supervisors and lose productivity and safety, or pay up for labour beyond what fixed-price bids assumed. About 62% of the workforce is unionised across about 450 agreements. That gives EMCOR flexibility, but it also brings contract-renewal and strike risk market by market.

EMCOR also participates in several multi-employer defined-benefit pension plans (2025 10-K). If a plan becomes underfunded or EMCOR withdraws from it, it can owe withdrawal liabilities that it does not control.

Calibration. The labour constraint is a high-probability limit on growth rather than a cause of losses. The pension tail is low probability but could be large in a specific plan.

8.4 Acquisition pace, pricing and write-downs

Mechanism. EMCOR has bought Miller Electric (its largest deal ever), Danforth, and at least 11 companies in 2026. Several were bought to move from warehousing or industrial work into data centers (Section 7). If those moves fail, or if acquired firms lose key people after earn-outs end, the acquired goodwill will need to be written down. The Q3 2026 write-down of an energy-efficiency contracting business in Building Services (Section 2) shows that acquired niches can lose their market. Competition for targets with data-center exposure is pushing prices up (Section 5), which reduces the margin for error on each deal.

Calibration. Moderate probability of further small write-downs. Low probability of a large one while data-center demand holds. The risk rises sharply if 8.1 occurs.

8.5 Weak non-data-center construction becoming a bidding war

Mechanism. The non-residential market outside data centers is flat to declining, and manufacturing construction is falling sharply (Section 6). Today, EMCOR's crews outside data centers are busy with healthcare, institutional and water work. If those sectors slow, EMCOR's crews and competitors' crews would chase fewer non-data-center jobs, and bids would get cheaper. A data-center slowdown would send regional competitors back into that same market at the same time, intensifying the pressure.

Calibration. Moderate probability, moderate impact. This is the usual late-cycle pattern for contractors.

8.6 Hyperscaler purchasing power

Mechanism. EMCOR's most important customers are among the most powerful buyers in the economy. They know contractor economics in detail, run competitive processes across a few qualified contractors and can push contract terms (GMP structures, liquidated damages, payment terms) toward less favourable positions over time. Guzzi's description of them as customers not willing to pay "one cent more than needed" (Section 4) is a warning as well as a compliment.

Calibration. High probability of gradual pressure on terms. The impact depends on how scarce qualified labour remains. Today labour scarcity favours EMCOR. In a slower market the balance would shift to the buyer.

8.7 Industrial Services exposure to refinery decisions and geopolitics

Mechanism. Turnaround volume depends on refiners' willingness to shut units. When oil supply is disrupted and refining margins are high, refiners defer maintenance. Guzzi said that would happen in Q4 2026 because of Middle East disruption (Q2 2026 call; see Section 9). Shop work is steadier, but field services can swing a small segment from profit to loss within a year, as it did in Q2 2025.

Calibration. High probability of volatility, low impact at group level given the segment's roughly 7% share.


9. Walk the Talk

Six results calls used:

  1. Q2 2025: 31 July 2025
  2. Q3 2025: 30 October 2025
  3. Q4 2025: 26 February 2026
  4. Q1 2026: 29 April 2026
  5. Q2 2026: 30 July 2026
  6. Q3 2026: 29 October 2026 (within 90 days of this report)
What was guidedWhenWhat happenedVerdict
FY2025 revenue and adjusted EPS ranges, tightened and raised through the yearQ2 2025 (raised EPS, narrowed revenue); Q3 2025 (tightened revenue, raised EPS again)FY2025 revenue and adjusted EPS both finished above the top of the Q3 2025 ranges (Q4 2025 release)Kept (beat)
Margins should be judged on a rolling 12-24 month average, within a stated bandQ3 2025 (Nalbandian)FY2025 adjusted operating margin landed at the top of the band. FY2026 guidance was then raised twice, and the Q3 2026 call guided above the bandKept, then exceeded
New-market startup losses in electrical would be a "one or two quarter headwind"Q3 2025 (Guzzi)Electrical margin recovered over the next three quarters and expanded strongly year on year in Q2 and Q3 2026Kept
UK sale for about US$255m, closing by year-end; Danforth closing in Q4 2025Q3 2025UK sale closed 1 December 2025; Danforth closed and its amortisation was built into 2026 guidanceKept
FY2026 revenue and EPS guidance, initial rangeQ4 2025Raised three times (Q1, Q2 and Q3 2026). The current revenue range sits well above the top of the original onePending (year-end), on track to beat
Construction margins within a stated rolling bandQ2 2025Electrical has stayed above the band in 2026, mechanical slightly below it as GMP and cost-plus mix grew. Combined construction is within the bandBroadly kept
Industrial Services turnarounds delayed from 2025 to late 2025 and 2026Q3 2025Segment revenue grew in Q1 and Q2 2026, then was flat in Q3 2026; Q4 2026 turnaround season flagged as weakPartly kept
New acquisitions to add H2 2026 revenue, with backlog amortising into 2027Q2 2026Deals closed early August. Q3 2026 showed meaningful acquired revenue in electrical, and management set out the 2027 backlog runoffKept to date, pending 2027

The margin call that proved accurate. At the low point of investor concern in Q3 2025, when electrical margins were falling because of startup costs in new markets and the stock was being questioned for it, the CFO set out a framework instead of a forecast:

"A rolling twelve to twenty-four month average is where we expect our margins to be." - Jason Nalbandian, Q3 2025 call, 30 October 2025

That was a cautious way to anchor expectations. The FY2025 result landed at the top of the stated band. By the Q3 2026 call, full-year 2026 margin guidance had moved above the band. Management under-promised and then delivered more than the framework.

The guidance pattern. FY2026 guidance was raised at every call after it was issued: first in Q1 2026 after a record bookings quarter, then in Q2 2026, then a third time on 29 October 2026. In FY2025, both revenue and adjusted EPS finished above the top of the last range given. Two years of this pattern points to deliberately conservative initial guidance, not a lucky run. Each raise came alongside record backlog, so it was earned by bookings already in hand rather than by hopes for future awards.

Where management was less precise. In Industrial Services, turnaround timing has repeatedly come in later than expected, and management has kept describing the work as pushed into a later period. Guzzi was candid about the latest delay:

"The turnaround season will not be as strong as it normally can be because... they have to keep open and keep producing oil and gas." - Tony Guzzi, Q2 2026 call, 30 July 2026

Q3 2026 Industrial Services revenue was flat, consistent with that warning.

The other blemish is the Q3 2026 write-down of the Building Services energy-efficiency business. It was small and non-cash, but it was not previewed on earlier calls.

Assessment. This management team does what it says and is consistently conservative in guidance. FY2025 finished above every range and FY2026 has been raised three times. Its long-range frameworks (margin bands, the growth gap over non-residential construction, the one-to-two-quarter startup cost) have held up. Misses are confined to the smallest and most cyclical segment and to one small acquired niche, and management flagged the Industrial Services problem before it showed up in results.


10. Ownership, Governance and Shareholder Friendliness

Part 1: Ownership and control. EMCOR is widely held and has no controlling shareholder. It has one class of common stock with one vote per share, and 44,440,278 shares outstanding at the 7 April 2026 record date (2026 proxy statement). The largest holders are index and institutional managers. The 2026 proxy cites a Schedule 13G/A filed by The Vanguard Group on 30 October 2025 reporting beneficial ownership of 5,332,636 shares, roughly 12% of shares then outstanding. Vanguard filed a further amendment on 26 March 2026 saying that after an internal realignment on 12 January 2026 it no longer reports beneficial ownership of shares held by various Vanguard subsidiaries and divisions. BlackRock also appears among the large holders, but its exact stake was not captured in the documents reviewed. There is no dual-class structure, pyramid, cross-holding or controlling family. Directors and executive officers own small stakes. For a minority shareholder, this structure means ownership and voting power line up one-for-one. Board control depends on director elections and the proxy-access right (3% held for three years can nominate up to 25% of the board). No single holder can block resolutions.

Part 2: Governance

  • Board. Nine of ten directors are independent. The CEO, Tony Guzzi, is also Chairman. The board offsets this with an independent Lead Director, M. Kevin McEvoy (2026 proxy).
  • Audit and related parties. Ernst & Young LLP is the auditor. No related-party transactions were identified in the materials reviewed.
  • Hedging and pledging. Directors and named executives are barred from hedging, margin accounts and pledging company stock.
  • Pay structure. Executive pay depends on measurable targets. The annual incentive is based on adjusted diluted EPS and an operating-cash-flow-to-operating-income ratio, with up to 20% for personal goals. Long-term pay combines restricted stock units that generally cliff-vest after three years with three-year cash awards tied to cumulative EPS (2026 proxy). The point worth noting is that every financial metric is EPS-based or cash-conversion-based, with no return-on-capital or relative-shareholder-return test. That structure rewards EPS growth from acquisitions and buybacks regardless of the price paid. For 2025, adjusted EPS and the cash-flow ratio both hit maximum payout.
  • Pay level and vote. CEO total compensation for 2025 was US$14.38 million, about 164 times the median employee (2026 proxy). The 2025 say-on-pay vote passed with more than 90% support.

Part 3: Capital returns

Dividends. The quarterly dividend was US$0.18 per share entering 2024. It was raised to US$0.25 from the April 2024 payment, giving US$0.93 paid in 2024. It was held at US$0.25 through 2025, giving US$1.00 paid in 2025. In December 2025 the board raised it 60% to US$0.40 from the January 2026 payment, giving US$1.60 paid across the four 2026 payments to date (2024 and 2025 10-Ks; Q2 and Q3 2026 releases). The dividend has grown each year, though it remains a small part of total cash returned. Buybacks do most of the work.

Buybacks and dilution. The board has added US$500 million to the open-ended repurchase programme three times recently: after the Q4 2024 results, in December 2025 and in October 2026. Executed buybacks, by period:

PeriodShares boughtCostSource
2023~0.8 million~US$128.7 million2023 10-K
2024~1.3 million~US$501.6 million2024 10-K
2025not captured~US$586.3 million2025 10-K
First nine months of 2026~0.6 million~US$455.2 millionQ3 2026 10-Q

From the programme's start through the end of 2024, EMCOR had bought back about 27.1 million shares for about US$2.39 billion. For the most recent ~90 days, MoatMap's database records one disclosure dated 29 October 2026: 200,000 shares at about US$949 per share, about US$190 million. That matches the third-quarter repurchases implied by the 10-Q. Shares outstanding fell from 47,047,265 at 31 December 2023 to 43,992,769 at 30 September 2026, a net reduction of about 6.5% in under three years. Buybacks have more than offset equity-compensation dilution, while EMCOR also funded record acquisitions over the same period.

Verdict: Returns Capital. The share count has shrunk about 6.5% in under three years through steady buybacks, and the dividend has risen every year, without slowing an acquisition programme that is the largest in the company's history.


11. Insider Activities

MoatMap's US Form 4 feed (current to 6 November 2026) holds no filings made before about 12 March 2026, so the months before that were read from the Form 4s on SEC EDGAR directly. The feed is the source from then on, and it was cross-checked against EDGAR for the last two weeks. EDGAR shows four Form 4s filed on 2 November 2026, which match the four 30 October 2026 entries in the MoatMap data: small dividend-equivalent credits on restricted units for Altmeyer, Guzzi, Mauricio and Nalbandian. No newer filings were found.

Excluded from the table as routine:

  • quarterly dividend-equivalent unit credits (up to 25 units each)
  • officers' restricted stock unit awards on 2 January 2026 (7,261 units to Guzzi, 1,507 each to Mauricio and Nalbandian, 327 to Lind) and the shares withheld for tax on 27 February 2026 (8,303 from Guzzi, 1,709 from Mauricio, 734 from Nalbandian, 233 from Lind)
  • the annual director equity grants on 4 June 2026, the date of the annual meeting (230-304 units each)
  • a gift of 1,377 shares by Maxine L. Mauricio on 17 March 2026
  • a sale of 0.32 of a share by Robert Peter Lind on 23 December 2025

Recent transactions (open-market and notable non-market, most recent first)

DateInsider (Name & Role)TypeSharesApprox ValueNotes
2026-09-01M. Kevin McEvoy (Lead Director)Purchase2~US$1,500Reported late, on 19 October 2026, with the June lot; too small to be a conviction purchase (SEC Form 4, filed 2026-10-19)
2026-06-17M. Kevin McEvoy (Lead Director)Purchase1~US$800Same late filing (SEC Form 4, filed 2026-10-19)
2026-06-17Carol P. Lowe (Director)Open-market sale950~US$0.80mReason not disclosed (SEC Form 4, 2026-06-17)
2026-05-11William P. Reid (Director)Open-market sale2,000~US$1.85mReason not disclosed (SEC Form 4, 2026-05-11)
2026-05-01John W. Altmeyer (Director)Open-market sale2,500~US$2.24mReason not disclosed (SEC Form 4, 2026-05-01)
2026-04-30Steven Schwarzwaelder (Director)Open-market sale5,602~US$5.01mReason not disclosed (SEC Form 4, 2026-04-30)
2026-04-30Jason R. Nalbandian (SVP, CFO & CAO)Open-market sale800~US$0.69mReason not disclosed (SEC Form 4, 2026-04-30)
2026-04-30Robert Peter Lind (VP & Controller)Open-market sale675~US$0.60mReason not disclosed (SEC Form 4, 2026-04-30)
2026-04-30Anthony J. Guzzi (Chairman, President & CEO)Gift3,000n/aNo sale proceeds (SEC Form 4, 2026-04-30)
2026-03-05Anthony J. Guzzi (Chairman, President & CEO)Open-market sale36,000~US$26.26mLargest sale in the window; under a Rule 10b5-1 plan dated 2 December 2025 (SEC Form 4, 2026-03-05)
2026-03-02Maxine L. Mauricio (CAO, EVP & General Counsel)Open-market sale3,357~US$2.47mReason not disclosed (SEC Form 4, 2026-03-02)

Buys: the signal. There were no meaningful open-market purchases in the last 12 months. McEvoy's two lots total three shares, which carries no signal. No officer or director made a conviction purchase.

Sells: the reasons. The largest sale was the CEO's. Guzzi sold 36,000 shares at an average of US$729.48 on 5 March 2026, about US$26.3 million, under a Rule 10b5-1 trading plan dated 2 December 2025. The 10-K describes the plan as covering up to 36,000 shares, and all of them sold on the day it took effect. The sale was about two-thirds of the roughly US$40 million sold in the window and about 17% of his direct holding, which it left at 170,299 shares including restricted units. Mauricio's sale came three days earlier, after the Q4 2025 results (26 February 2026). Five of the other six sales fell in a ten-business-day window right after the Q1 2026 results (29 April 2026), the usual opening of the post-earnings trading window. That timing points to routine diversification by directors and officers who hold stock built up from years of equity grants, after a large share-price rise. None of those seven filings disclosed a 10b5-1 plan, an estate sale or a tax-related reason, so for each of them the reason is not disclosed. Their sizes are small against total shares: the largest, Schwarzwaelder's about US$5 million, was 0.01% of shares outstanding. Just before the window, director Robin A. Walker-Lee sold 1,325 shares, about US$0.9 million, on 3 November 2025.

Net assessment. Insiders were net sellers in the open market: eight sales worth about US$40 million against a token purchase. About two-thirds of that was the CEO's plan sale. The rest came from four outside directors and three officers, mostly after one earnings window and modest in size. Every director and officer meets the company's stock-ownership guidelines (2026 proxy). The pattern is consistent with routine diversification, not a change in view, but the CEO did cut his direct holding by about 17%, and no insider was willing to buy at these levels either. Read: neutral to mild concern.


12. Scenarios

Bull case. The AI buildout moves from first-generation training campuses to sustained multi-campus programmes, and the cloud-to-AI mix reaches 50/50 earlier than management expects. Each building carries far more piping and power, and EMCOR's prefab shops capture more of it. Hyperscalers that already share multi-year plans with EMCOR award follow-on phases on proven campuses rather than re-bidding them. Contracting with a few partners nationally becomes the norm, and EMCOR is one of a very small group able to provide electrical, mechanical and fire work in more than a dozen markets. The 2026 electrical acquisitions in Austin, Chicago, Ohio and Wisconsin all move into data-center work. Healthcare, university and Florida water projects keep non-data-center crews busy. In two to three years EMCOR is a larger, more fabrication-heavy national trade partner to hyperscalers, with a much bigger services base maintaining the systems it installed. Its multi-market supervisory bench, which no rival roll-up can reproduce quickly, strengthens the moat from narrow toward wide.

Base case. Management delivers roughly what it has guided. The doubled data-center backlog converts through 2027 with the usual one- or two-quarter startup costs in new markets. The GMP and cost-plus shift keeps mechanical margins slightly below past peaks while profit dollars grow, the trade-off management has said it prefers. EMCOR keeps adding one or two data-center markets a year and buys several regional firms annually. A few acquisitions underperform and lead to small write-downs like the Q3 2026 energy-efficiency charge. Building Services grows at a single-digit rate on retrofits and controls. Industrial Services moves with refinery timing. Guidance keeps being set conservatively and beaten, consistent with the record in Section 9. In two to three years EMCOR is clearly larger and more concentrated in data centers. It remains a well-run, conservatively guided consolidator whose results depend on the hyperscaler capex cycle it has chosen to lean into.

Bear case. Risks 8.1 and 8.5 compound. A hyperscaler capex reset (an AI monetisation disappointment, a power-supply wall, or a credit tightening affecting the developers) means follow-on phases stop being awarded. Backlog keeps revenue going for a few quarters while new bookings shrink. Regional competitors pushed out of data centers flood the weak non-data-center market, and bids for healthcare, institutional and commercial work get much cheaper just as EMCOR needs them. Risk 8.4 then appears: acquired electrical firms bought to move into data centers have nothing to move into, and goodwill is written down. Risk 8.3 adds the cost of stranded supervisors in rural markets entered for single campuses. Under 8.6, hyperscalers use the slack to tighten GMP terms. EMCOR remains solvent and still returns cash, because its balance sheet and service base are sound. But it is a smaller, more competitive and less differentiated contractor than its 2026 backlog suggested.


Sources:

Charts

Want a report like this on a company you hold?

This one was free to read, like every finished report in the library. Search any of 25,000+ companies. A new Deep Dive costs 50 Feathers, and the price shows again before anything is spent.

Done reading EMCOR Group, Inc.?

Here's what to check out next.

Get the monthly Best Stocks email, and alerts when a stock you track moves.

Sign up free →

EMCOR Group, Inc. (EME) Deep Dive — AI Research Report

EMCOR Group, Inc. (EME) — Executive Summary

When a new hospital, chip plant or AI data center is built, the concrete and steel are the easy part. What makes the building work is inside the walls.

This is the executive summary of a 10,000+ word (about 45 min read) AI-generated research report. The full report covers business segments, earnings transcript analysis, management credibility, competitive landscape, valuation, risks, and bull/bear scenarios.

Frequently Asked Questions

What does EMCOR Group, Inc.’s (EME) deep dive cover?
MoatMap’s deep dive on EMCOR Group, Inc. (EME) is an AI-generated equity research report covering business segments, earnings transcript analysis, management credibility, competitive moat, peer comparison, valuation, risks, and bull/bear scenarios. The full report is approximately 10,000 words (about 45 minutes of reading).
Who writes MoatMap deep dives?
An AI model writes each deep dive from the company's filings and earnings calls, its peers' financials and the wider economic picture. No person edits it, and it does not update itself: every report carries the date it was written, and an older one says so. It is research, not personal investment advice.