S14.1 · Real Estate & Built Environment
US homebuilding and land development, roughly $900-950B annualized, where AI is compressing entitlement and back-office costs faster than field construction.
Residential homebuilding is the development, entitlement and construction of US housing communities — roughly $900-950B in annualized private residential construction put-in-place in 2025 (US Census Bureau), with the public homebuilders (D.R. Horton, Lennar, PulteGroup) generating combined revenue near $100B. Volume has run flat to down since 2023 under elevated mortgage rates, but the rate cycle is not the variable to watch. The variable to watch is where the large builders' advantage actually lives — purchasing scale and back-office overhead — because AI is already compressing the entitlement-processing and administrative layer that advantage rests on, and it compresses for regional builders too.
Flat-to-down activity through 2023-2025 traces directly to mortgage rates suppressing transaction volume. The National Association of Home Builders forecasts a low-single-digit recovery in 2026-2027, resting on rate relief plus a structural underbuilding gap estimated at 3-5 million units — a demand floor, not a growth engine. Revenue and land supply sit in the Sun Belt — Texas, Florida, North Carolina, Arizona, Georgia — while entitlement friction is heaviest in coastal metros, where the gap between demand and deliverable lots is widest. The top 10 public builders take roughly 35-40% of closings; the large regional and private tail beneath them continues to consolidate.
The value chain runs land developers and lot-option specialists, then builders, then trade contractors and building-products suppliers. The economics are a hybrid worth keeping separate. Land banking and lot options behave like a real estate asset business — capital-intensive, cyclical, priced off rate and cap-rate conditions — while the home sale itself is a one-time transaction with no recurring tail. Gross margins run 15-25% and move with rates more than with anything management controls.
Of the segment's natural adjacencies — residential construction and remodeling, and building-products supply — integration into either defends margin rather than adding revenue: a builder that controls trade capacity and materials supply is buying cycle-time and margin protection in a labor-scarce market. Land banking is the expansionary use of capital, a bet on future entitlement value.
AI's measurable effect so far sits around construction, not in it. Entitlement and permitting document review and construction-draw administration are being compressed by tools already in use, and the money question is who that compression pays. The answer is the smaller operator: scale builders have held their edge over regional and private builders largely through purchasing scale and back-office overhead, and cheap AI-driven documentation work shrinks the second of those for everyone at once.
Field construction moves on a different clock. Skilled-trades labor — electrical and mechanical journeymen above all — is the binding constraint regardless of demand conditions, and physical installation stays human-labor-bound over a 5-10 year horizon. Efficiency gains land in pre-construction and administration first; how homes get physically built changes later, if at all.