The 2026 AC Replacement Cost Index: What Can Actually Be Measured

This is a research page, not a buying guide. It states its method, publishes the figures that have a federal source behind them, labels the arithmetic we did ourselves, and says plainly which numbers are contractor-reported market estimates with no authority behind them at all. If what you want is a price by tonnage and efficiency tier so you can size up a quote, that lives on our AC replacement cost guide — this page deliberately does not duplicate it.

Why “average AC replacement cost” is a number that is never actually measured

Search the phrase and you will get a confident dollar figure inside three seconds. Follow it back and the trail goes somewhere uncomfortable. The large home-services cost guides quote each other and quote contractor blogs; the contractor blogs quote the cost guides. Nothing at the bottom of the pile is a sample.

That is not an accusation of dishonesty. It is a structural fact about this market. The federal agencies that publish energy data publish energy data: the Energy Information Administration tracks what a kilowatt-hour and a therm cost, the Environmental Protection Agency tracks refrigerant regulation, the Bureau of Labor Statistics tracks what manufacturers charge for equipment at the factory gate. None of them tracks what a contractor charges to put a system in your house, because nobody collects that. There is no HVAC equivalent of the Consumer Price Index for installed work.

The consequence for a homeowner is specific: a precise “average” is the least trustworthy number on any cost page, because precision is the one thing the underlying evidence cannot support. A wide range with its basis stated is worth more than a narrow one with a footnote.

Our method

What we collected, and from where

Two classes of input, kept strictly apart throughout this page.

• Sourced inputs. Federal and utility publications with a URL and a date: DOE’s Federal Energy Management Program acquisition guidance for central air conditioners, air-source heat pumps and furnaces (guidance updated December 2024); EIA’s Electric Power Monthly and its California natural gas series; EIA’s Residential Energy Consumption Survey reporting on household air-conditioning spending; EPA’s HFC allowance allocation and technology-transition pages; IRS Fact Sheet FS-2025-05; AHRI’s residential central air conditioner efficiency standards page; and the City of Sacramento’s published HVAC change-out permit fee structure. Every one was fetched and status-checked on the dates given in the sources list at the foot of this page.
• Contractor-reported market estimates. The installed price ranges. These come from the general market conversation — cost guides, quoted jobs, trade discussion, and they have no authority behind them. They appear here labelled, never cited, and published wide rather than precise.

Collection window

Sourced inputs were resolution-checked and their figures extracted in August 2026. Each carries its own publication date in the sources list, and several are considerably older than this page: DOE’s FEMP guidance was updated in December 2024 and computes in 2023 dollars; the EIA household-spending figures are 2015 survey data published in 2018. Those vintages are stated wherever the numbers appear, because a stale figure presented as current is the specific failure this page exists to avoid.

What counts as a “replacement”

Throughout this page, a central AC replacement means: a new outdoor condenser, a matched indoor evaporator coil, standard labor, reuse of sound existing ductwork and line set, and startup and test. That definition matters more than any number attached to it, because it is the definition two quotes silently disagree on.

What we excluded, and why

Excluded from every base figure on this page: ductwork, line set replacement, electrical or panel work, the permit, disposal of the old equipment, crane or rooftop rigging, California Title 24 HERS verification, and a filter drier upgrade. Each is a real cost and several are near-universal, but they vary enough between houses that folding them into a headline range makes the range meaningless. They are itemized further down instead.

Also excluded: any regional multiplier. The site’s own tooling and its own cost guide have historically disagreed about what California adds, and a single statewide uplift is structurally wrong in any case — the inland Central Valley and the coastal Bay Area are different labor markets. We would rather publish no multiplier than an unfounded one.

What we can publish, and what we cannot

An index of this kind normally opens with a median, an interquartile range and a long tail. We are not publishing those, and the reason is the finding. A median requires a sample. We do not have a priced sample of actual homeowner invoices, and a median synthesised from cost-guide ranges would be a weighted average of the same unmeasured sources this page just finished criticising — dressed in the language of statistics, which would make it worse rather than better.

What can be published is a span, labelled for what it is. Contractor-reported market rates put a full central AC changeout, scoped as defined above, somewhere between $3,200 and $14,000+ — from a small baseline-efficiency system on existing ducts at one end to a large premium variable-speed system at the other. That is a genuinely wide band and the width is the information: it tells you that the equipment decision is not what separates the two ends. Scope is, which is also why brand choice moves a quote far less than most buyers expect, a point we take up in our guide to AC brands. For what the install itself involves and how long it takes, see new AC installation cost.

The index: DOE’s efficiency math, re-priced at California energy costs

Here is the part of this page that is genuinely new, and it comes entirely from federal inputs.

DOE’s Federal Energy Management Program publishes, for each major appliance class and by climate region, the annual and lifetime energy cost at each efficiency level, and — most usefully — the maximum price premium at which the more efficient model still pays for itself. California sits in DOE’s Hot-Dry/Southwest region for cooling equipment and in its Southern region for furnaces. Those tables are the closest thing to a federal answer to “is the efficiency upsell worth it,” and they are rarely published.

Central air conditioner, DOE Hot-Dry/Southwest region, at DOE’s own assumptions (25-year lifetime, 2023 dollars):

DOE’s cost-effective premium ceiling for this region: $869. In plain terms, paying more than about $869 extra for the ENERGY STAR unit over the less-efficient one stops paying for itself over the equipment’s life, on DOE’s assumptions. Best-available lifetime savings in this region: $3,154.

Air-source heat pump, same region (15-year lifetime, 2023 dollars):

Cost-effective premium ceiling: $728. Best-available lifetime savings: $5,177.

Gas furnace, DOE Southern region — note that the ENERGY STAR level here is 90% AFUE, not the 95% used in the Northern region (21-year lifetime, 2023 dollars):

Cost-effective premium ceiling: $549. Best-available lifetime savings: $947.

The caveat that changes everything, and the arithmetic we did about it

DOE computes those dollars at 11.0 cents per kilowatt-hour — the average price paid at federal facilities as of July 2024, and at $0.961 per therm for gas. California households do not pay anything like that.

The following three numbers are our arithmetic, not a DOE result, and we are labelling them as such. Scaling DOE’s premium ceilings by the ratio of California energy prices to DOE’s assumptions gives:

• Central AC, ENERGY STAR over less-efficient: roughly $2,600 rather than $869.
• Air-source heat pump: roughly $2,200 rather than $728.
• Gas furnace, Southern region: roughly $1,150–$1,250 rather than $549.

The method is a linear scaling of the energy-cost component by the price ratio. It does not re-run DOE’s discounting, does not adjust the 2023-dollar base, and assumes California usage matches DOE’s regional profile, so treat these as an order-of-magnitude correction rather than a recomputed federal figure. What they establish is directional and important: at California electricity prices, the efficiency premium that pays for itself is roughly three times what DOE’s published table shows. Nobody else has published DOE’s own tables re-priced for the most expensive electricity market in the continental United States, and we would rather show the arithmetic than assert the conclusion.

One honest counterweight sits alongside it. EIA’s Residential Energy Consumption Survey put average household spending on air conditioning at about $265 a year nationally, $525 in the hot-humid Southeast, and roughly $60 in the temperate marine region along the West Coast, where nearly half of households did not use air conditioning at all. Those are 2015 survey figures published in 2018 and electricity prices have risen materially since. But the shape holds: on the mild California coast, cooling spend is small enough that no efficiency premium pays back quickly. The inland Central Valley runs far more cooling hours and is a different case entirely, and EIA does not publish a Central Valley figure, so we are not putting a number on it.

For how the efficiency tiers actually translate into a purchase, see what SEER2 rating to buy and our explainer on what SEER2 means.

The one federal price series that does exist: equipment at the factory gate

Everything above this point concerns energy cost. There is exactly one federal series that tracks the price of the hardware itself, and it is rarely published in this niche: the Bureau of Labor Statistics Producer Price Index for industry 333415.

Two series, taken every March so the comparison is like-for-like. The index values are BLS actuals; the percentage changes are our arithmetic on those published values.

Over the seven years from March 2019 to March 2026, air conditioner prices at the factory gate rose 53.9% and furnaces 58.8%.

The shape of that curve is the story, not the headline percentage. There were two surges, not one. The first was 2021–22, when air conditioners jumped 15.3% and furnaces 21.4% in a single year. Then came a nearly flat stretch: air conditioners rose just 2.2% and then 1.3%, and furnace prices actually fell 2.0% in the year to March 2025. Then a sharp turn upward into 2026. A homeowner who last priced a system in 2024 and assumes the market has been drifting gently upward since is working from the flat part of a curve that has ended.

What we will not tell you is why. The A2L refrigerant transition is the obvious candidate and it is chronologically plausible, but the index does not say why prices moved and neither can we. “Equipment prices rose 7.8%” is a sourced fact. “Equipment prices rose 7.8% because of the refrigerant transition” is a claim with nothing behind it, and this page does not make claims like that.

Two dating notes that belong with the numbers. The series run through July 2026, but April through July 2026 are flagged preliminary and BLS is carrying earlier values forward pending revision, so the last firm value in each series is March 2026, and every percentage above uses it. And the air conditioner series is defined as unitary air conditioners excluding heat pumps; it says nothing about heat pump equipment prices, and we have not found the correct series for those.

What moves the number most

Ranked by how much they move a real quote, all of these are contractor-reported market estimates unless noted.

• Ductwork. The single largest swing factor. Full replacement runs roughly $3,500–$12,000 for a typical single-story home, with large multi-story or all-sheet-metal jobs reaching about $14,000, or roughly $25–$55 per linear foot installed. Whether your ducts survive the changeout can double the project.
• Efficiency tier. Stepping up a tier adds roughly $400–$1,400 at mid-range, $1,000–$2,400 at high efficiency, and $1,400–$3,000+ for premium variable-speed. Compare that against the premium ceilings above — that comparison is the whole point of this page.
• Tonnage and correct sizing. Equipment is selected on rated capacity at design conditions, which is ACCA Manual J for the load and Manual S for the equipment selection, not a rule-of-thumb multiplier. Oversizing is a comfort and humidity problem as well as a price one; see how to size an AC with Manual J.
• Electrical scope. A dedicated circuit, outdoor disconnect and whip runs roughly $250–$1,000 combined. A full service panel upgrade from 100A to 200A is a different and much larger job at roughly $1,500–$4,500. Those two get conflated constantly; they are not the same line.
• Line set. Replacement runs roughly $300–$800. Whether it can be reused depends on condition, sizing for the new equipment, and refrigerant compatibility.
• Disposal and refrigerant recovery. Roughly $100–$500, and the requirement behind it is not optional. EPA Section 608 prohibits venting refrigerant during service or disposal and requires certified handling. This belongs inside the base price as an itemized line, not in an “optional add-ons” list.
• Permit and inspection. No national figure is defensible and we are not publishing one. Fees are set by each city or county in its own schedule. The City of Sacramento, for example, publishes a dedicated residential HVAC change-out fee page naming five separate charges — building permit fee, general plan fee, green building fee, technology fee, and a city operations tax charged when a California licensed contractor holds the permit — with the building permit fee itself valuation-based rather than flat. Look up your own jurisdiction; a permitted, inspected install is also what most incentive programs require.
• California Title 24 and HERS verification. Third-party HERS rater verification of duct leakage and refrigerant charge applies to many California duct alterations and change-outs. Out-of-state cost guides omit this line entirely, which is one concrete reason a national average misprices a California job. We do not publish a combined permit-plus-HERS figure because the two are separate charges and merging them produced one of the contradictions this index was built to retire.

What changed since 2025

The end of R-410A manufacturing, and what it did and did not do

Three separate things get blurred together constantly, so here they are separated. One: manufacture and import of new residential R-410A equipment ended January 1, 2025. Two: EPA’s final rule of May 26, 2026 removed the installation compliance date entirely for systems whose specified components were all manufactured or imported before that date — meaning remaining stock has no installation deadline. Three: servicing and recharging an installed R-410A system remains legal indefinitely, with no end date.

The mechanism behind rising refrigerant prices is the strongest sourced fact on this subject: EPA’s HFC allowance allocation drops from 60% of baseline (2024–2028) to 30% (2029–2033). The legal supply is cut in half again at the start of 2029. That is why refrigerant costs climb, and it is a schedule rather than a rumour. Per-pound refrigerant pricing itself, however, remains genuinely unsourceable — no authority publishes it, so we treat any per-pound figure as a market estimate. Our refrigerant recharge cost page handles that in detail.

A2L refrigerants and what they mean in a new system

New equipment ships on R-454B or R-32, which are A2L refrigerants — mildly flammable. If you suspect a refrigerant leak, kill ignition sources and ventilate the space, and get a licensed technician in; homeowners cannot legally buy most refrigerants and cannot service a sealed system. One practical consequence worth knowing before you pay for a repair: A2L systems carry a refrigerant sensor, and when it trips the system runs the blower and locks out cooling. “Fan runs, no cold air” on a new system can be a working safety feature rather than a fault — that is a diagnosis, not a reason to authorize a repair on the spot.

The loss of the federal credit

Under Public Law 119-21, enacted July 4, 2025, the Section 25C energy efficient home improvement credit was terminated for property placed in service after December 31, 2025, and the Section 25D residential clean energy credit for expenditures made after the same date (IRS Fact Sheet FS-2025-05). There is no federal residential tax credit for a new air conditioner, heat pump or furnace in 2026, air-source or geothermal. For anyone comparing a 2025 quote to a 2026 one, that is an effective price increase of up to $2,000 on a qualifying heat pump, and it is the largest single year-over-year change in the economics of this purchase. State and utility rebates are the remaining incentives, and they are funding-limited and change often.

What this index cannot tell you

Stated in full, because a study that lists no limitations has not been examined.

1. It cannot tell you the median installed price. There is no measured sample behind any installed price on this page, ours included. Nothing here should be read as “what you should pay.”
2. It cannot give you a regional multiplier. None is published, and a single California figure would misprice half the state.
3. Its equipment price trend is not an installed price trend. The Producer Price Index above is factory-gate data. Labor rates, contractor margin, permit fees and regional differences all move independently of it, and none of them is measured here. A 7.8% rise in what manufacturers charge does not translate into a 7.8% rise in your quote. The index also carries no regional information whatsoever, so no California claim can be built on it, and it excludes heat pumps entirely.
4. Its most recent months are provisional. BLS flags April through July 2026 as preliminary in both series and carries earlier values forward pending revision, so this edition stops at the March 2026 firm value. We were also unable to confirm the publication date of the underlying BLS news release: the BLS website returns an automated-access block, and the figures here were pulled from the sanctioned public API on August 20, 2026 rather than read off a dated release. That is stated rather than papered over.
5. Its energy figures carry vintages. DOE’s tables compute in 2023 dollars from December 2024 guidance; the household-spending figures are 2015 survey data published in 2018. Our re-pricing is a linear scaling, not a re-run of DOE’s model.
6. It cannot price your permit, your crane, or your ductwork. Permit fees are municipal. We publish no figure at all for crane or rooftop rigging, because there is no defensible one and inventing one would be the exact defect this page was built to correct.

One more, in the same spirit: the commonly repeated “an AC lasts 12 to 15 years” has no published source behind it, while DOE’s own economic analysis assumes 18 years for a central air conditioner, 15 for an air-source heat pump and 21 for a furnace, and uses 25 years for the Hot-Dry/Southwest AC table above. Field lives are genuinely shorter than design lives, but a shorter assumed life systematically pushes a repair-or-replace calculation toward “replace”, which is the recommendation that costs a homeowner the most.

How to use this before you get quotes

1. Fix the scope before you compare prices. Write down whether each quote includes ductwork, line set, electrical, permit, disposal and — in California — HERS verification. Two numbers are not comparable until this is done, and it is the step most buyers skip.
2. Ask for the load calculation. Manual J for the load, Manual S for the equipment selection. A contractor who sizes off the old unit’s nameplate is reproducing whatever mistake was made last time.
3. Price the efficiency step against the ceiling, not against the brochure. Take the actual dollar difference between the tiers you are quoted and compare it to the premium ceilings above — DOE’s published figure, or our California-scaled version if you are on California electricity rates. That single comparison is worth more than any efficiency marketing you will read.
4. Check your own utility for a rebate before you sign. With the federal credit gone, utility programs are the remaining money, and several are larger than the credit was. They are funding-limited and require specific equipment and a participating contractor, so confirm eligibility before the equipment is ordered rather than after.
5. Treat any precise “average” you find — including on competitor pages — as an estimate with no sample behind it. That includes the ranges on this page. We publish them wide for that reason.

Methodology appendix and reuse terms

How to read any figure on this page. Every number here falls into exactly one of three classes, and the class is stated wherever the number appears:

• Sourced — from a federal, regulatory or utility publication with a URL and a date. Quotable as fact with the citation. The DOE efficiency tables, the EIA energy prices, the EPA refrigerant schedule, the IRS credit status and the AHRI efficiency standards are all in this class.
• Derived — our arithmetic from sourced inputs. Two things on this page are derived and both are labelled where they appear: the California re-pricing of DOE’s premium ceilings, and the year-over-year percentage changes calculated from the BLS index values (the index values themselves are BLS actuals).
• Contractor-reported market estimate — no authority behind it. Every installed price on this page. Published as a wide range, never cited to a source, never presented as measurement.

Citation line for press and researchers. HVACFixPro, “The 2026 AC Replacement Cost Index,” August 2026, hvacfixpro.com/ac-replacement-cost-index-2026/. The tables above may be reproduced with attribution and a link. If you reproduce the California-scaled premium ceilings, please carry the label with them — they are our arithmetic on DOE inputs, not a DOE publication. Our sourcing and correction practice is described on our editorial standards page; if you find an error here, we would rather hear it than not.

What would make the next edition better, named openly: a re-pull of the BLS series once the April–July 2026 preliminary values are revised, and the correct BLS series for heat pump equipment, which we have not identified; a compiled table of published municipal HVAC change-out permit fees across Northern California jurisdictions, which is real primary-source data nobody has aggregated; and a genuine sample of homeowner invoices with stated geography and scope. The last is the only thing that would support a median, and a hundred invoices is the threshold, not a dozen.

FAQ