Your Teardown Is Holding $3,280 in Lumber. The Excavator Will Never Know.
The excavator arrives on a Tuesday. By Friday the 1952 ranch house is a twelve-ton pile of fir framing, oak flooring, and one extremely heavy cast-iron tub, all of it riding to the landfill at sixty dollars a ton. Somewhere in that pile are two thousand square feet of old-growth fir framing, cut from forests that no longer exist and milled to dimensions no lumberyard stocks, and all of it will be buried under six feet of trash by the weekend, which is the part nobody puts on the bid sheet. Nobody inventoried what was inside the walls before the bucket came down, and on residential teardowns, nobody ever does.
Demolition produces more than 90 percent of America's construction and demolition debris, according to the EPA. The agency's last national estimate put the total at 600 million tons a year, with nearly 145 million tons landfilled. That estimate is from 2018, and no federal agency has published an update since; with construction spending up more than 40 percent in the interim, a stretch covering the post-pandemic building boom and the Infrastructure Investment and Jobs Act, the real number is certainly and substantially higher (Waste Advantage Magazine). We are burying houses faster than ever, and we have stopped bothering to count, which means every local government writing disposal policy today is doing it against a number from two administrations ago, in a construction economy that no longer exists.
145 million tons. Construction and demolition debris landfilled in the US in 2018, the last year the EPA measured. Demolition, not construction, generated over 90 percent of it.
The Bid Sheet Is Missing a Line
Deconstruction, the careful unbuilding of a house to salvage its materials, rarely wins the bid. Not because the economics fail, but because the bid sheet has no line for what the old house is worth.
Researchers at the University of Florida deconstructed six wood-frame homes and ran the numbers without flinching. Deconstruction cost about 21 percent more than demolition up front; then the researchers priced the salvaged lumber, doors, and fixtures at a fraction of retail, and the net result came in roughly 37 percent cheaper than demolition (Guy & McLendon, 2000). A nonprofit in Bellingham, Washington documented a similar story on a 1,292-square-foot house: 97 percent of the building diverted from landfill, $2,647 in salvaged material, a little over two dollars per square foot (the RE Store case study). At Cornell, Felix Heisel's team took apart a 4,500-square-foot house in five days and found the salvage revenue nearly erased the labor premium, a result that should embarrass every contractor who claims deconstruction can never compete, though it is worth noting that Ithaca has both a strong reuse market and a university willing to treat a teardown as a research project, which is not exactly a replicable business model (Yale Environment 360).
So why does the excavator win every time? Those studies priced the salvage after the fact. At bid time, the homeowner faces a firm demolition quote against a deconstruction quote built on guesswork. Certainty beats virtue on every spreadsheet, which is exactly why this is an information problem wearing a cost problem's clothes.
A Camera That Prices Your Walls
That gap is what a new wave of AI tools is trying to close. Autodesk Research has built a system that scans interior walls with multimodal sensors and identifies recoverable materials behind the finishes, hitting 95 percent accuracy detecting studs in controlled tests. In a drywall-recovery case study, one test wall section held $121 in reusable panels against $102 in disposal costs, plus about 40 kilograms of avoided CO2. Small numbers, but the method is the point: the salvage gets priced before the demolition rather than lamented after it. What the scan really sells is not the inventory but the timing, because a salvage estimate delivered before the demolition contract is signed can change which bid wins, while the same estimate delivered a week later is just an elegy for lumber.
Other teams are working the adjacent links in the chain. Researchers at Monash University trained deep-learning models on the first real-world image dataset of contaminated demolition wood and reached 91 percent accuracy spotting paint, chemical, and metal contamination from ordinary photos (published in Resources, Conservation and Recycling). Contamination is the quiet reason most demo wood gets landfilled, since no reuse yard will buy lumber that might be carrying lead paint and no crew can afford to hand-inspect every board that comes off a teardown.
In Basel, Switzerland, civic digital databases already track reusable building materials and match them with buyers before demolition begins, which is the low-tech version of exactly what the AI scanners promise: a market that can see the goods before the building comes down. In Ukraine, an AI trained on 2,518 annotated images of damaged buildings now sorts rubble into reuse, recycle, and disposal streams, and flags asbestos for special handling (Dezeen). Closer to home, an Oakland company called Urban Machine is building robots that pull nails out of salvaged lumber. Anyone who has ever denailed a 2x4 by hand will recognize that as the single most tedious step in the entire salvage economy.
~$4,000. Estimated recoverable value in a typical 2,000-square-foot teardown: about $3,280 in salvaged material plus roughly $700 in avoided landfill tipping fees, with the full arithmetic laid out in the section below.
Do the Math on Your Own Teardown
Take a typical 2,000-square-foot teardown. The Bellingham crew weighed its house at 12.3 pounds of building per square foot, which scales to about 12.3 tons for ours. Landfill tipping fees for construction debris run $40 to $70 a ton across much of the country and are climbing (Wilmot Inc.). At a $60 midpoint, burying the whole house costs roughly $740 in disposal alone. Divert 97 percent of it, as the Bellingham crew did, and most of that $740 comes back.
Then the salvage: the Florida researchers valued recovered material at an average of $1.64 per square foot, which on 2,000 square feet comes to $3,280 in year-2000 dollars. Real money, even before inflation adjusts it upward. Add the avoided disposal and the old house is carrying roughly $4,000 in recoverable value.
Now price the information: a 360-degree camera walkthrough of a house takes an afternoon, and the analysis is software. Suppose an AI pre-demolition audit cost $500, a figure I am inferring from adjacent computer-vision capture services rather than quoting from any vendor, as flagged in the limitations section below. Fifteen percent of $3,280 is about $490, so the audit pays for itself if it lifts the realized salvage value by just 15 percent, before counting avoided disposal or the tax deduction for donated materials. It does not need to be perfect; it needs to beat a guess, because a guess is what it replaces.
The Honest Objections
None of this repeals the labor math. A Canadian government study found labor costs dominate the deconstruction equation so completely that a 10 percent swing in wages moves the project total more than twice as much as a 10 percent swing in salvage value. In high-wage metros the deconstruction premium has reached 78 percent over demolition, which is brutal for anyone selling deconstruction on economics alone, because wages only rise while the price of used 2x4s answers to a local salvage market that might not exist within a hundred miles of the job site. An AI can inventory your walls flawlessly and still lose to the excavator if nobody within fifty miles buys used 2x4s.
Most of this technology is also younger than it sounds. Ninety-five percent stud detection happened under controlled conditions in an Autodesk lab, not in a 1920s balloon-frame house with three unpermitted remodels and a century of paint. The 91 percent contamination detector was validated on a curated dataset. And the $1.64-per-square-foot Florida figure assumed salvaged materials sold at up to half of retail with zero storage, transport, or sales costs booked against them, which describes no actual business. No storage fees. No sales staff. No buyback.
Then there is the hazmat wall: in a pre-1978 home, lead paint and asbestos can make the framing legally unsalvageable regardless of what the camera sees, and no algorithm repeals that. Check first.
What This Analysis Could Not Prove
A few honest boundaries. No vendor publishes per-home pricing for an AI pre-demolition audit, so the $500 audit figure is an inference, not a quote. The Florida $1.64 figure is from the year 2000, covers six houses in one market, and the materials were never actually sold. Tipping fees vary wildly by region, from around $30 a ton in parts of the Southeast to over $100 on the coasts, which moves the avoided-disposal math substantially. The Autodesk and Monash results are controlled-condition research, not field deployments on ordinary housing stock. And I deliberately did not quote current reclaimed-lumber retail prices, since I could not verify them against a primary source, which means the salvage side of this ledger should be read as a documented historical range rather than a promise about any particular teardown.
What to Do If You Are Tearing Down
If a teardown is in your future, five moves capture most of the value.
First, get the deconstruction bid alongside the demolition bid, and demand an itemized salvage estimate. A contractor who cannot itemize is pricing on vibes. Second, ask for the tipping-fee line item in writing, and ask who keeps the avoided-disposal savings. At $60 to $100 a ton, every diverted ton is money, and it should be yours. Third, look into donating salvaged material through a 501(c)(3) reuse store; the tax deduction can exceed what the material would fetch in cash, especially for high-income homeowners. Fourth, get the hazmat survey before anything else on a pre-1978 home, since lead and asbestos decide what is legally salvageable. Fifth, plan reclaimed lumber for non-structural uses like fencing, sheds, or interior cladding. Most codes will not accept it structurally without re-grading, so confirm with your building department before designing around it. None of these moves requires waiting for the AI tools to mature, which is convenient, because the scanners are still mostly research projects and your teardown is scheduled for October.
The excavator will always be faster. Speed was never the question. Every teardown is a sealed box with a price tag inside, and for a century the industry has priced the box at exactly the cost of smashing it, which is not thrift but a failure of imagination wearing a hard hat. The question is whether we keep mistaking the absence of information for the absence of value, and keep burying four thousand dollars a house because nobody thought to look inside the walls first.