# Why PE Pays 14x EBITDA for HVAC Data Center Contractors Canonical: https://granular.to/blog/pe-pays-14x-ebitda-hvac-data-center-contractors Published: 2026-06-24 Updated: 2026-06-24 Author: Trey Category: Operator's view Tags: hvac, custom-software, ai-agents, operations > Hyperscaler capex topping $830 billion in 2026 has split the HVAC private equity multiple chart in two. Mid-market commercial HVAC shops with documented data center backlog now clear 12-14x EBITDA while generic peers trade at 7-10x, leaving owners three strategic paths: build the operating system for capacity ramp, defend commercial service, or sell into a roll-up. > **TL;DR.** PE-backed HVAC platforms paid 10.9x EBITDA on average in 2025 (Capstone Partners), and Apollo wrote Apex Service Partners up to ~20x at a $10B valuation in May 2026. The middle of the chart has split. Generic mid-market commercial HVAC trades at 7-10x. Contractors with documented data center backlog and qualified field labor clear 12-14x. Comfort Systems USA's same-store backlog doubled from $5.99B to $11.58B in twelve months, with 45% of 2025 revenue from technology customers. If you run a $40-80M commercial HVAC shop, you have three real paths: chase data centers, defend commercial service, or sell into a roll-up. The HVAC private equity multiple chart broke in two over the last twelve months. The sector average sits at 10.9x EBITDA across 149 transactions in [Capstone Partners' 2025 HVAC M&A review](https://www.adastraequity.com/ebitda-multiples/hvac-business). Apollo's May 2026 minority investment in Apex Service Partners marked the platform tier at roughly 20x EBITDA on a $10 billion valuation. The middle of the chart now splits on one question: does this contractor have data center exposure? Mid-market commercial HVAC contractors with documented hyperscaler backlog and qualified field labor are clearing 12-14x EBITDA today. The same revenue and margin profile without that exposure clears 7-10x. The four-to-five turn spread tracks one thing: where data center capex is going. ## The $830 billion capex story (and the 55 cents) [TrendForce's May 2026 forecast](https://www.trendforce.com/presscenter/news/20260506-13033.html) puts 2026 capex across the top nine global CSPs at roughly $830 billion, with growth revised from 61% to 79%. Microsoft raised its 2026 outlook to $190 billion (130% YoY). AWS is on pace for $230 billion+. Google guided to $180-190B, Meta $125-145B. Rich Miller's [datacenterrichness.substack.com](https://datacenterrichness.substack.com/p/hyperscalers-plan-630-billion-in) puts Big Four 2026 capex around $630 billion, up 62% from 2025. Two numbers do the work for an HVAC owner. Dell'Oro Group pegs total data center capex at roughly $600 billion for 2026, up 25% YoY. Per a [Buildermuse market read](https://buildermuse.com/commercial/hyperscaler-data-center-capex-2026/), 55 to 65 cents of every data center construction dollar goes to mechanical and electrical scope. Roughly 60%. The Dodge Momentum Index runs in single-digit-percent swings quarter over quarter; hyperscaler capex is running at 60%+ annual growth, allocated through earnings calls 18 months before the permit hits the portal. ## What Comfort Systems' backlog actually tells you Comfort Systems USA (NYSE: FIX) is the public-market comp every mid-market HVAC owner should be reading every quarter. The company reported [2025 revenue of $9.1 billion and adjusted EBITDA of $1.45 billion](https://www.businesswire.com/news/home/20260219524928/en/Comfort-Systems-USA-Reports-Fourth-Quarter-and-Full-Year-2025-Results), with same-store backlog rising from $5.99 billion to $11.58 billion in twelve months. Roughly doubled. The most important number sits in the [10-K filing](https://www.stocktitan.net/sec-filings/FIX/10-k-comfort-systems-usa-inc-files-annual-report-380f0303a17f.html): technology customers represented 45% of 2025 revenue, up from below 25% three years ago. Comfort Systems is not a data center pure-play; it runs 50 operating units across 190 locations. But almost half of every dollar in 2025 was tied to hyperscaler and colocation work, and that mix is what is taking the public-market EV/EBITDA multiple into the mid-teens. ![Macro view of large-diameter chilled water piping and primary loop manifolds inside a hyperscale data center mechanical room, showing the scale of HVAC infrastructure that drives mid-market contractor valuations](/images/blog/pe-pays-14x-ebitda-hvac-data-center-contractors-chilled-water-loop.jpg) That is the visibility premium at work. A 10-year capex pipeline turns a mid-cycle mechanical contractor into a backlog-visible compounder. Capital markets pay differently for backlog visibility than they do for bid pipeline. The same logic now extends down the chart. A regional $50M HVAC shop with two qualified data center campuses in backlog gets priced by acquirers on the same visibility premium. ## Your three strategic paths The split in the multiple chart has narrowed the strategic decision for a $40-80M commercial HVAC contractor to three honest options. ### Path A: Build the operating system for data center capacity The work is sitting there. The constraint is operating capacity: estimating discipline for hyperscaler-spec submittal packs, a field labor pipeline that can absorb a 30% headcount ramp without quality collapse, and prefab or modular spool capacity that lets you commit to schedule certainty on a $40M mechanical package. The shops winning today already have operating systems that standardize submittals, run a stretch labor pipeline, prove safety and quality metrics on demand, and run lookahead schedules at the package level. The multiple math: a $50M shop with $5M EBITDA at 7x clears $35M. The same shop with two hyperscaler campuses in backlog and a documented prefab operation at 13x clears $65M. The 18-month operating system investment is the difference between those two exits. The [parallel electrical playbook for data center work](/blog/mid-market-electrical-contractor-data-center-playbook) walks the same stack from the EC side. ### Path B: Defend commercial service If you cannot ramp skilled trades on a 12-18 month timeline, defend the service-contract base. Mid-market commercial HVAC contractors with service-contract revenue mix above 40% hold a multiple floor at 8-10x because recurring revenue is what PE acquirers actually price. You will not get the data center premium. You will get a defensible multiple with no execution risk. Service-agreement pricing has improved 20-30% across the trade in the last 18 months ([CT Acquisitions' 2026 HVAC valuation reference](https://ctacquisitions.com/guides/hvac-business-valuation/)). A $50M shop with $5M EBITDA at 8.5x clears $42.5M. Less upside than Path A, predictable, runnable with the team you already have. ### Path C: Sell to a roll-up There are 27+ active PE platforms in HVAC in 2026 (per [CT Acquisitions' platform map](https://ctacquisitions.com/guides/private-equity-hvac-2026/)). Apex Service Partners (Apollo, $10B), Wrench Group (Leonard Green / Apax), Sila Services (Morgan Stanley Capital Partners, $1.7B), Champions Group (Blackstone, $2.5B at 18.5x), Redwood Services (Altas / Sun Capital, $1.1B), and Service Logic (Bain Capital and Mubadala) lead the platform tier. Apex alone closed approximately 60 add-on acquisitions in 2025. Add-on shops trade at 4-8x in the sub-$2M EBITDA tier and 6-11x in the $2M+ tier with strong recurring mix. Platform-level recaps clear 17-20x. The deal stack is typically 50-70% cash, 10-15% earnout, and 15-30% rollover equity that lives or dies on the platform's second exit. If you have a sub-scale operation, a soft service-contract base, and a tired ownership group, Path C is rational. If you have any of the three Path A ingredients, selling now is leaving 3-5 turns of multiple on the table. ![Mechanical contractor's prefab assembly floor with workers fabricating modular pipe spools on jig tables, illustrating the off-site capacity that drives data center bid wins for mid-market HVAC shops](/images/blog/pe-pays-14x-ebitda-hvac-data-center-contractors-prefab-fabrication.jpg) ## The binding constraint isn't technology. It's the operating system. The trap most mid-market HVAC owners walk into is treating data center work as a technology problem. CAD packages, BIM coordination, design-assist tools are commoditized. Comfort Systems is not winning because their software stack is better than yours. Comfort Systems is winning because they can show a hyperscaler a labor pipeline of 22,700 employees, a documented safety and quality history, modular fabrication that pre-builds 40% of mechanical scope off-site, and a service organization that takes the building back at substantial completion. Hyperscaler bids reward four things: schedule certainty, mod/prefab capacity, qualified labor count, and service-contract follow-on. You do not need to match Comfort's 22,700 headcount. You need to be the regional shop that can take a $20-40M secondary mechanical package on a $1B campus, run it without dropping schedule or quality, and convert into the service contract for the next decade. That is a viable $80-120M business, and it is what the 12-14x multiple is pricing. The operating system has five pieces that have to actually work: - An estimating package that produces hyperscaler-spec submittals in under four weeks from RFP - A labor pipeline that can stretch 25-30% above current headcount in 12 months without collapsing first-time quality - Prefab capacity that pre-builds at least 30% of mechanical scope off-site, measured in spool-feet per labor hour - A service-contract follow-on motion that captures recurring revenue on every install completed - A field execution layer that puts safety, QA, and schedule lookahead in one place with weekly GC visibility Each one is buildable in 18 months. None are buildable in six. ## What changes for you in 2026 The window is open and compressing. Comfort Systems, EMCOR, and Limbach are taking the headline data center mega-contracts at the top. Mid-market shops are picking up the secondary packages, campus retrofits, and service follow-on. Service Logic and Apex are still adding bolt-ons at 6-10x; the ones with data center exposure are clearing the middle of the chart at 12-14x. If you wait, the spread compresses. The HVAC PE platforms know this; that is why they are running 60 add-on acquisitions a year. If you build, you exit at the top of the spread. If you defend service, you sit on a solid floor. If you sell now, you take a multiple that is real today but will not be the high. ## FAQ **What multiple should a $50M commercial HVAC contractor expect in 2026?** No data center exposure, typical 10-12% EBITDA margin: expect 7-10x EBITDA. With a documented hyperscaler backlog of $20M+ and a qualified labor pipeline: expect 12-14x. Service-contract revenue mix above 40% without data center exposure: expect 8-10x with strong floor support. Apex / Apollo set the platform ceiling at roughly 20x in May 2026; that is not the comp for a regional $50M shop. **What counts as "data center exposure" for valuation purposes?** Two things, both verifiable. First, documented backlog from a named hyperscaler or top-five colocation operator (Equinix, Digital Realty, QTS, CyrusOne, Aligned). Second, evidence of operational capacity to execute: prefab capacity above 20% of scope, a labor pipeline that has grown 15-20% in the last 12 months, and at least one substantially completed hyperscaler campus on the resume. Bid pipeline alone does not count. **How fast can a mid-market shop build data center capacity?** 12-18 months for the operating system pieces (estimating, prefab, project controls, service-contract motion). 18-24 months for the labor pipeline to support a 25-30% headcount ramp. The shops moving fastest have a dedicated data center operations lead reporting to the COO, a separate estimating bench for hyperscaler work, and a prefab investment in the $2-5M range. **Should I sell now or wait to ramp?** Depends on your starting point. With one hyperscaler campus in execution and 15+ percent service-contract revenue mix, the wait pays: two more years of execution moves you from 9-10x to 12-14x on a higher EBITDA base. With neither and an ownership group over 60, the cleaner trade is selling into a roll-up with strong platform economics and rollover equity. The worst path is half-ramping. ## The 18-month decision The data center capex flow is not going to wait. The hyperscalers have already pre-committed roughly $630 billion in 2026 capex on the Big Four alone. The window where a $50M shop can build the operating system to clear 12-14x is open for the next 18 months, and it closes when capacity gets absorbed. The contractors building those operating systems right now are the ones who will set the floor for the 2028 multiple chart. The ones treating data center work as a technology problem will not. At Granular, we build the operating systems mid-market HVAC contractors need to stand up data center capacity: estimating, prefab discipline, labor pipeline tracking, service-contract follow-on, without three years of Procore custom configuration. Most of our HVAC clients run between $40M and $80M and have at least one hyperscaler bid in flight. If you want a straight read on which of the three paths you can actually execute, [book 30 minutes with us](/). One call, no pitch deck. --- ## Keep Reading - **[The Mid-Market Electrical Contractor's Data Center Playbook](/blog/mid-market-electrical-contractor-data-center-playbook)**: The parallel playbook for $40M MEP and specialty electrical contractors building hyperscaler campuses. - **[What Hyperscaler Capex Means for Mid-Market MEP](/blog/hyperscaler-capex-mid-market-mep)**: The broader Money Flows piece on where the data center buildout is reshaping mechanical, electrical, and plumbing demand.