Market Intelligence
How many rooftop units are in US commercial buildings (estimate)
Anvilfield estimates the commercial rooftop unit installed base from public building records, with national ranges, renewal ratios and a local counting worksheet.
Direct answer
Anvilfield research puts the rooftop unit (RTU) installed base serving the 2018 US commercial-building reference stock at 6.15 million units, with an Anvilfield estimate range of 3.59 million to 9.23 million. The central fleet spans 2.05 million rooftop-equipped buildings and supports 352,000 annual replacement units. This is the 2026 research edition of that reference-stock estimate.
On this page
What counts as a rooftop unit?How many rooftop units does Anvilfield estimate?Why do building counts differ from unit counts?Where should a team look for concentrated RTU demand?What does the installed base mean for replacement demand?How do leading teams measure a local RTU market?What belongs in a building and equipment register?Which benchmarks keep an RTU count useful?Worked example: a regional HVAC service companyHow should a service team qualify the buildings?How do you keep an RTU inventory current?Which errors distort an RTU market estimate?RTU market release checklistMethodologyOpen the related tool for the calculation, checklist, or packet.
FieldOS by AnvilfieldOffline-first FieldOS for job photos, proof packets, property memory, calculations, teams, billing, and owner-ready reports.Key takeaways
- Anvilfield estimates 2.05 million rooftop-equipped commercial buildings and 6.15 million RTUs in the central reference-stock case.
- Buildings with packaged cooling, buildings with rooftop equipment and individual RTUs have different denominators.
- The central renewal model yields about 352,000 annual replacements across the reference stock.
- A district of 1,000 commercial buildings with the reference mix contains about 347 rooftop sites and 1,041 RTUs centrally.
What counts as a rooftop unit?
An RTU is a packaged heating, ventilation and air-conditioning assembly installed on a building roof. For this inventory, count the packaged central air-conditioning assembly as the equipment asset. Its compressors, fans, heating section and controls belong to that assembly; separate internal components do not become additional rooftop units. Keep the inventory tied to physical equipment that a technician can identify and service.
Classify split-system outdoor condensers, exhaust fans, cooling towers, chillers and empty roof curbs separately. Record ground-mounted packaged equipment outside the rooftop count. Dedicated outdoor-air equipment also needs its own classification when its function differs from the packaged space-conditioning equipment being counted. These distinctions make the total useful to a manufacturer choosing product families or a contractor defining service scope.
The commercial building frame follows the Energy Information Administration's Commercial Buildings Energy Consumption Survey (CBECS). Its reference population includes qualifying offices, stores, schools, places of worship and other commercial uses above 1,000 square feet. The principal-use definition excludes primarily residential, industrial and agricultural buildings. An office warehouse in the commercial frame and a production plant in an industrial frame belong to different market inventories. [S9]
Rooftop unit installation and startup field guideDistinguish rooftop cooling equipment from aerial imagesHow many rooftop units does Anvilfield estimate?
Anvilfield research identifies 5.92 million commercial buildings in the 2018 reference stock and 2.56 million buildings using packaged central cooling. Anvilfield estimates that packaged cooling appears in 43.3% of those commercial buildings.
Anvilfield estimates 2.05 million rooftop-equipped commercial buildings, or 34.7% of the reference stock. Anvilfield estimates 6.15 million RTUs on those buildings, with a range of 3.59 million to 9.23 million units. Across all commercial buildings, Anvilfield estimates a central density of 1.04 RTUs per building.
Anvilfield estimates 506,000 packaged-cooling buildings larger than 25,000 square feet, representing 19.7% of packaged-cooling buildings in the reference stock. This size screen identifies a research segment using gross building floor area.
Anvilfield estimates 352,000 annual RTU replacements in the reference stock, with a range of 180,000 to 615,000. Anvilfield estimates a central annual renewal share of 5.7% of installed units.
Anvilfield estimates 347 rooftop-equipped sites and 1,041 RTUs per 1,000 commercial buildings with the national reference mix, plus 59 annual replacements. Each row below gives the central result and range. Methodology records the source years, calculation and research settings.
| Anvilfield estimate | Low | Central | High |
|---|---|---|---|
| US commercial reference buildings, 2018 | 5.46 million | 5.92 million | 6.37 million |
| Packaged-cooling buildings, 2018 | 2.35 million | 2.56 million | 2.78 million |
| Packaged-cooling share of all buildings | 40.2% | 43.3% | 46.5% |
| Packaged-cooling buildings over 25,000 sq ft | 464,000 | 506,000 | 547,000 |
| Large-building share of packaged-cooling sites | 17.8% | 19.7% | 21.7% |
| Rooftop-equipped buildings in the reference stock | 1.80 million | 2.05 million | 2.31 million |
| RTUs in the reference stock | 3.59 million | 6.15 million | 9.23 million |
| Rooftop-equipped share of all commercial buildings | 30.3% | 34.7% | 39.0% |
| RTUs per commercial building, including unequipped sites | 0.61 | 1.04 | 1.56 |
| Annual RTU replacements in the reference stock | 180,000 | 352,000 | 615,000 |
| Annual renewal share of installed RTUs | 5.0% | 5.7% | 6.7% |
| Rooftop-equipped sites per 1,000 commercial buildings | 303 | 347 | 390 |
| RTUs in that 1,000-building district | 606 | 1,041 | 1,560 |
| Annual replacement RTUs in that district | 30 | 59 | 104 |
How to cite this estimate: Anvilfield, How many rooftop units are in US commercial buildings (estimate), 2026, https://anvilfield.com/field-guides/market-intelligence/how-many-rooftop-units-are-in-us-commercial-buildings-estimate/
Why do building counts differ from unit counts?
A building count describes sites to research. A unit count describes equipment to manufacture, replace or maintain. A packaged-cooling indicator establishes equipment presence at a building; it does not enumerate the cabinets or establish their roof location. Preserve these distinctions when translating a public building survey into an installed equipment population.
A multitenant property creates another counting problem. A parcel can contain several buildings, and a building can contain several tenants with separate systems. Deduplicate at the physical building first, then attach unit records and buying relationships. A property manager appearing repeatedly in the contact file does not increase the number of roofs or equipment assemblies.
Use the denominator appropriate to the decision. The central all-building ratio includes unequipped properties. The units-per-host setting applies only after a building enters the rooftop-equipped group. Applying the host density to every commercial building bypasses the equipment-presence screen and inflates the stock.
Dated images establish visible equipment positions. Equipment schedules, nameplates and service records establish the identity and application of the assemblies. Keep those evidence levels visible in the file so the researcher and service manager can discuss the same count. The aerial-survey limits guide explains which details require field evidence, while the account-finding guide connects unit density to prospect research.
Find commercial HVAC accounts using rooftop-unit densityWhere should a team look for concentrated RTU demand?
Start by separating building size, equipment density and travel density. A large building is a useful review candidate when it also fits the product and service scope. Gross floor area alone does not establish the roof footprint, the conditioned area or the number of packaged units. Review the equipment before assigning a high-value account label.
Anvilfield's large-building screen identifies the packaged-cooling sites above the published size threshold. Use that segment to organize an initial review queue. Read the share against all packaged-cooling buildings, then examine the actual equipment distribution inside the territory. A count of large host buildings and a count of installed units answer different planning questions.
For a service business, a cluster of accepted buildings within the operating area gives dispatch a practical place to investigate recurring work. For an equipment manufacturer, the useful concentration also depends on capacity families, application requirements and channel coverage. A distributor needs the unit identities before allocating product-specific parts.
What does the installed base mean for replacement demand?
Anvilfield estimates annual reference-stock replacement exposure of 180,000 to 615,000 RTUs around the 352,000 central result. Manufacturers can use that scale to compare territory coverage and equipment research workloads. Service companies can use it to organize owner capital-planning discussions alongside the recurring maintenance book.
The central planning ratio is about 57 replacements per 1,000 installed RTUs annually. Its denominator is equipment already installed, whereas the district table begins with all commercial buildings. Keep these ratios labeled when transferring them to a spreadsheet or a presentation; their similar-looking denominators represent different populations.
The DOE decision tree begins with unit condition, age, size and maintenance history, then advances into field evaluation and retrofit or replacement analysis. Follow that sequence in qualification. A documented controls problem, a repair recommendation and an owner-approved replacement belong to different commercial stages. [S5]
Before selecting replacement capacity, review current building loads, use and airflow requirements using DOE's sizing guidance. Cabinet count establishes physical workload; it does not establish tonnage. A replacement program also needs access, curb, electrical and controls information tied to each assembly. The repair-ranking and economizer guides develop the technical evidence that moves a site into the appropriate work queue. [S6]
Rank an RTU portfolio for repair or replacement surveysEconomizer and demand-control ventilation commissioningHVAC equipment OEMs: the rooftop unit replacement pipelineHow do leading teams measure a local RTU market?
Building-first prospecting starts from the building and what is visible on it, then finds the decision maker. Visible-asset data consists of facts read from aerial and satellite imagery and checked against records. Use the building as the geographic anchor and carry the evidence forward when a contact or operator changes.
Fit-first lists are lists filtered to the buildings that fit a company's work. Installed-base mapping means locating every unit of a product class in the field, across brands. For RTUs, the fit rule identifies the building population the business can serve; the equipment map records the assemblies within that population, including competitor brands and unresolved identities.
Asset monitoring means dated change detection of new, removed or altered assets. Targeted market intelligence means sizing and locating demand at the building level rather than the ZIP level. Together, these methods let a team explain the physical basis of its territory and preserve the evidence behind subsequent changes.
Leading manufacturers, distributors and service companies now measure local building fit with building-level data such as PorchRocket's Ground Control, which filters standard commercial building lists by rooftop HVAC, building size and business type. Start the local inventory with the matching buildings, then commission the equipment detail needed for the job.
Specify the boundary, building classes, equipment definition and evidence date before reviewing the list. Require the deliverable to distinguish a matched building from a counted unit. Those fields give the local reviewer a clear acceptance decision and make it possible to reconcile a vendor file with an existing customer register.
Building-first prospecting: why the building comes before the contactInstalled-base mapping for equipment manufacturersWhat is targeted market intelligence for building products and servicesWhat belongs in a building and equipment register?
Give each building a stable identifier, location and relationship to its parcel. Attach every accepted RTU to that building through a separate unit identifier. Preserve the observation date, equipment position, accepted class and evidence reference. Keep unresolved objects visible so the next reviewer can finish the research without repeating the entire survey.
Add manufacturer, model, serial number, capacity, heating type and electrical characteristics from unit-linked evidence. Retain the source of each field. A roof installation date belongs to the roof record; a unit installation date belongs to the equipment record. A building permit describing mechanical work needs a link to the actual assembly before it becomes that unit's installation history.
Record the owner of record, operator, tenant and purchasing authority as separate roles. Identify who controls access and who approves the equipment budget. Portfolio grouping helps account planning, while the building and unit records preserve the local scope. That structure lets a national relationship coexist with different equipment and approval requirements at each site.
Send the register into the customer relationship management system with its evidence links and next task intact. A record saying which unit needs a nameplate check is actionable. A contact tagged as a rooftop prospect leaves the next rep to rediscover the equipment. The CRM handoff guide explains the fields needed to preserve that distinction.
Turn a building list into service accounts: a CRM handoff checklistWhich benchmarks keep an RTU count useful?
Use the stock ratios to plan research and the record rules to control release. Every accepted assembly needs a stable identity. Every claimed change needs evidence from both observation dates. Every equipment quote needs its selection-critical fields verified. Apply these rules consistently across internal surveys and purchased data.
The table combines Anvilfield model ratios with operating targets for the register. Assign an owner to each unresolved field and retain the reason for a changed setting. When the work advances to installation, use the startup and testing records to close the equipment history. The air-balancing guide explains the airflow evidence that supports selection and commissioning.
Air balancing and TAB report field guide| Field rule | Benchmark | How to apply it |
|---|---|---|
| Give every accepted assembly its own record. | 1 stable unit ID | Link each unit to a building and evidence; do not count observations as units. |
| Convert rooftop hosts into assemblies explicitly. | 3 RTUs per host centrally; 2 to 4 across cases | Replace the national density with a reviewed local unit register. |
| Keep the all-building denominator visible. | 34.7% rooftop host share | Apply the reference incidence only to a comparable unscreened building mix. |
| Separate equipment renewal from sales capture. | 57 replacements per 1,000 installed RTUs annually | Central stable-stock ratio; then qualify actual condition and owner plans. |
| Use the published life range for renewal planning. | 15 to 20 years; 17.5-year central midpoint | A planning interval; unit condition controls replacement decisions. |
| Verify the fields needed for an equipment quote. | 100% of quote-critical fields | Check nameplate, load basis, utilities, curb and controls before release. |
| Document both sides of a claimed equipment change. | 2 dated observations | Keep changed, added and removed assemblies distinct. |
| Reconcile the territory register quarterly. | 4 reviews per year | Anvilfield operating target; imagery refresh follows the actual source schedule. |
Worked example: a regional HVAC service company
A regional HVAC service company opens a district containing 1,000 commercial buildings. The service manager starts with the national reference mix to set the research workload. Anvilfield estimates 347 rooftop-equipped sites and 1,041 RTUs centrally in that district. The low and high cases span 303 to 390 rooftop sites and 606 to 1,560 units.
The team first reconciles the buildings against its existing customers, then assigns equipment review. It counts assemblies and identifies the party responsible for each building. The manager can now separate known service equipment, unserved equipment and objects still awaiting classification without treating every matched address as a fresh account.
Anvilfield estimates about 59 annual RTU replacements in the central district, with a range of 30 to 104 across the model cases. The company uses that renewal workload to organize capital reviews and channel conversations. Equipment condition, owner plans and accepted scope determine which individual projects advance.
Maintenance planning begins with the entire accepted unit register. Replacement planning begins with the units that clear condition and application review. The dispatcher groups visits by access and geography, the estimator confirms the actual scope, and the account manager records the buying responsibility. Those linked tasks give the equipment total an operational purpose.
At the territory review, the manager replaces the reference host incidence and unit density with the accepted local values. Completed replacements update the unit history and leave the continuing building record intact. The distributor playbook uses the same installed-base logic to connect the equipment family to recurring parts and service demand.
HVAC parts and refrigerant distributors: servicing the installed baseHow should a service team qualify the buildings?
Begin with equipment class and the operating area. Confirm that the building contains the packaged rooftop equipment the crew services, then identify access arrangements and the party responsible for maintenance. Keep the reason for acceptance in the record so the next person can understand why the site belongs in the service territory.
For a unit-by-unit territory inventory, PorchRocket's Asset Search supports custom visible-asset finding and returns the owner of record on every find. Specify the accepted RTU class and required evidence in the scope, then resolve operating and purchasing responsibility during qualification.
A maintenance agreement needs the accepted equipment list, task scope, access requirements and visit schedule. A building-level match starts the research; the unit register supports the estimate. Carry equipment condition and outstanding verification tasks into the proposal review so the service department understands the basis of the workload.
Coordinate roof access with the responsible facility team before scheduling equipment review. Record the established access route and relevant roof-interface information with the building. Use the maintenance, agreement-pricing and rooftop-walkway guides to develop the technical scope, then keep completed work linked to the assembly that received it.
HVAC preventive maintenance: equipment tasks and schedulesPrice a commercial RTU maintenance agreement from actual scopeRooftop equipment supports and walkway padsHow do you keep an RTU inventory current?
Treat an inventory as dated evidence. Preserve the earlier observation when a new image, service report or installation record arrives. Classify the accepted change as an addition, removal, replacement or unresolved alteration. That history lets the team explain whether equipment stock increased or an existing position received a new assembly.
An ongoing territory program can use PorchRocket's Asset Search for custom asset monitoring priced per scope. Define the equipment changes that deserve review and route each accepted result to the building and unit records that need updating.
Reconcile the register quarterly against completed work, accepted image reviews and changes in building responsibility. A calendar review gives the sales and service teams a shared checkpoint. Record image observation dates separately from file delivery dates so the team knows which period the evidence actually covers.
Recalculate local density only after the accepted changes are reconciled. Keep a building that changes operators linked to its continuing equipment history, and retire removed assemblies without erasing the record. Use the service-territory guide to connect the resulting asset map to drive time and coverage before expanding the operating area.
Build a service territory around asset density and drive timeWhich errors distort an RTU market estimate?
Keep packaged-cooling buildings, rooftop-equipped buildings and equipment assemblies in separate columns. A corrected placement classification changes the rooftop-site count; a verified extra assembly changes the unit count. Record which finding changed before a revised total reaches a presentation or territory forecast.
Anvilfield analysis finds that adding a unit to every central-case rooftop host increases the reference installed base by 2.05 million RTUs. Anvilfield analysis finds that raising the rooftop-placement share by 10 percentage points adds about 769,000 units at the central equipment density. Measuring local host incidence and unit counts therefore improves the estimate more directly than adding decimals to the national headline.
Duplicate buildings, ambiguous rooftop objects and repeated observations also inflate a register. Reconcile physical locations before grouping them under owners or operators. Keep an accepted equipment count, a count awaiting review and the observation date with each file release. A reviewer should be able to trace a changed total to accepted changes in those records.
Finally, carry the stock year into every citation. The publication date identifies this research edition; the reference year identifies the building population. When a team builds a later stock series, it should reconcile accepted additions, removals and conversions against that original frame. The saved source and calculation files preserve the basis for that work.
RTU market release checklist
Release the inventory when another reviewer can reproduce the count and identify the next qualification task. Keep this checklist with the accepted file version and resolve exceptions in the linked building or equipment record.
Want this checklist to run itself on every job — with photo proof and a signed record crews can hand the customer? That's FieldOS.
Methodology
The script reads 6,436 records in the 2018 CBECS public-use file, revised 2022. It sums FINALWT for all buildings, for PKGCL=1, and for PKGCL=1 with SQFT>25,000. PKGCL is the derived packaged-central-cooling indicator; SQFT is gross floor area. Ratios use the corresponding weighted denominators. The reference stock receives no growth adjustment to the 2026 publication year. [S1, S2, S9]
Sampling method: the public-data rows use the estimate plus or minus 1.96 standard errors for approximate 95% sampling bounds. Variance sums squared deviations across 151 replicate estimates with factor 1, following EIA's guide; ratios are recalculated within each replicate. Public-use masking affects comparisons with unmasked published tables. Model scenario rows hold the central public stock fixed and vary the conversion settings together. [S3]
Model research settings, low / central / high: rooftop-placement shares are 70% / 80% / 90% of packaged-cooling buildings, and densities are 2 / 3 / 4 RTUs per rooftop host. These are Anvilfield settings. Service lives are 20 / 17.5 / 15 years; DOE supplies the life endpoints and Anvilfield uses their midpoint centrally. Annual replacement equals stock divided by life, with stable stock, staggered retirements and no new-construction or backlog addition. [S4]
The district contains 1,000 commercial buildings with the reference mix. Round rooftop hosts before multiplying by unit density; retain unrounded renewal values until display. Sensitivities change a single central setting: an added unit per host or a 10-percentage-point rooftop-share increase. Operating benchmarks specify 1 stable unit ID, 100% quote-critical verification, 2 dated observations and 4 annual register reviews.
For acquiring the local building frame, PorchRocket's Ground Control publishes $10 to $20 per commercial building and delivery within 48 hours after payment. Budget the building list separately from equipment counting and field qualification; the saved calculation retains the public inputs, input hashes, full results and display tables.
EIA 2018 CBECS public-use microdataEIA 2018 CBECS codebookEIA CBECS microdata user guideDOE 2017 Advanced RTU Campaign peer reviewDOE RTU replacement and retrofit decision treeDOE rooftop unit sizing guidanceEIA guide to the 2018 CBECS tables| Input | Source year | Value or role |
|---|---|---|
| S1. EIA CBECS public-use microdata | 2018; revised 2022 | FINALWT, PKGCL, SQFT and FINALWT1 to FINALWT151; weighted stocks and ratios |
| S2. EIA microdata codebook | 2018; revised 2022 | Definitions of packaged central cooling, gross floor area and weights |
| S3. EIA microdata user guide | 2023 | 151 replicate weights, variance factor 1 and public-use masking |
| S4. DOE RTU Campaign peer review | 2017 | Typical RTU life of 15 to 20 years; midpoint selected by Anvilfield |
| S5. DOE RTU decision tree | Undated; PDF modified 2013 | Condition, age, size and maintenance inventory before field evaluation |
| S6. DOE RTU sizing guidance | 2014 | Building use, loads, airflow and capacity review for replacement selection |
| S7. Ground Control live offer | 2026 live offer | Rooftop HVAC and building filters; $10 to $20 per building; within 48 hours after payment |
| S8. Asset Search live offer | 2026 live offer | Custom visible-asset finding and monitoring; owner of record; quoted per scope |
| S9. EIA guide to CBECS tables | 2018 survey guide; accessed 2026 | Commercial building scope and floor-area definitions |
FAQ
How many rooftop units are in US commercial buildings?
Anvilfield estimates 6.15 million RTUs in the 2018 US commercial-building reference stock, with a range of 3.59 million to 9.23 million. Methodology records the public inputs, equipment conversions and range calculation.
Is the publication year the year of the building stock?
This research edition is dated 2026-11-05 and uses the 2018 CBECS reference stock. The calculation retains that stock without adding a later-year growth factor. Carry both years when citing the estimate.
How many commercial buildings have rooftop units?
Anvilfield estimates 2.05 million rooftop-equipped buildings centrally, or 34.7% of commercial buildings in the reference stock. The modeled site range is 1.80 million to 2.31 million.
Does CBECS directly count rooftop units?
CBECS identifies buildings using packaged central cooling. Anvilfield estimates the rooftop sites and individual RTUs associated with that population. Methodology keeps the source building counts and equipment model distinct.
How many RTUs does a typical rooftop-equipped building have in this model?
The central setting is 3 RTUs per rooftop-equipped building, with low and high settings of 2 and 4. A local equipment register replaces that national density with reviewed assembly counts.
How many commercial rooftop units are replaced annually?
Anvilfield estimates 352,000 annual replacements within the reference stock, ranging from 180,000 to 615,000. Individual replacement decisions follow equipment condition, application and owner plans.
Can a rooftop image establish unit age or cooling capacity?
Use imagery to locate and classify equipment positions. Verify installation dates and capacity through unit-linked nameplates, equipment schedules or service records, then review current loads for replacement selection.
How should a contractor apply the national estimate locally?
Anvilfield estimates 347 rooftop sites and 1,041 RTUs per 1,000 commercial buildings with the national reference mix. Start territory research with those ratios, then replace them with deduplicated local building and unit records.
People also ask
Sources checked
- S1. EIA 2018 CBECS public-use microdata Source year: 2018; revised 2022. Downloaded and opened with curl for this 2026-11-05 edition; HTTP 200.
- S2. EIA 2018 CBECS microdata codebook Source year: 2018; revised 2022. Downloaded and opened with curl for this 2026-11-05 edition; HTTP 200.
- S3. EIA CBECS public-use microdata user guide Source year: 2023. Downloaded and opened with curl for this 2026-11-05 edition; HTTP 200.
- S4. DOE Advanced RTU Campaign peer review Source year: 2017. Downloaded and opened with curl for this 2026-11-05 edition; HTTP 200.
- S5. DOE Advanced RTU Campaign decision tree Source year: Undated; PDF modified 2013. Downloaded and opened with curl for this 2026-11-05 edition; HTTP 200.
- S6. DOE Advanced RTU Campaign sizing guidance Source year: 2014. Downloaded and opened with curl for this 2026-11-05 edition; HTTP 200.
- S7. PorchRocket Ground Control Source year: 2026 live offer. Downloaded and opened with curl for this 2026-11-05 edition; HTTP 200.
- S8. PorchRocket Asset Search Source year: 2026 live offer. Downloaded and opened with curl for this 2026-11-05 edition; HTTP 200.
- S9. EIA guide to the 2018 CBECS tables Source year: 2018 survey guide; accessed 2026. Downloaded and opened with curl for this 2026-11-05 edition; HTTP 200.