US multifamily completions fell from roughly 608,000 units in 2024 — the highest annual total since 1986, per the Census Bureau's Survey of Construction as tabulated by the National Association of Home Builders — to about 484,000 units in 2025, per Census Bureau construction data, marking the turn of the record supply wave. Where those units landed, and where the shrinking pipeline stops landing, reshapes the rent map for 2026. This is analysis of public construction data, not investment advice.
Completions are the supply that actually competes for tenants. Announcements, permits, and even starts matter only insofar as they become deliveries, and the delivery schedule is now falling — with a lag built into the market's response.
What Did the Record Supply Wave Actually Deliver?
The wave crested in 2024. Builders completed about 608,000 multifamily units that year, per the NAHB's July 2025 tabulation of the Census Bureau's Survey of Construction, the strongest year since 1986. The following year totaled roughly 484,000 completions, per Census Bureau construction data — still elevated by historical standards, but a clear step down from the peak. The national rental vacancy rate, which reached 7.1 percent in the third quarter of 2025 per the Census Bureau's Housing Vacancies and Homeownership Survey, absorbed the front edge of that wave.
The geography of the deliveries was concentrated. Sun Belt metros — the Texas Triangle, Florida, Atlanta, Phoenix, and the Carolinas — took the largest shares of the 2022-2024 starts that became 2024-2026 deliveries, per the same construction series. That concentration is why national averages blended flat-to-falling rents in supply-heavy Southern submarkets with still-rising rents in coastal and Midwest markets.
Why Do Completions Fall Before Anyone Decides Anything?
The pipeline arithmetic is fixed years ahead. A multifamily project that starts construction in 2023 completes in 2025 or 2026; nothing about 2026 demand conditions changes that schedule. Multifamily starts declined from their 2022-2023 peak as financing costs rose, per Census new residential construction releases, and those missing starts become missing deliveries with a two-to-three-year delay.
This is the mechanical core of the current period: 2024's record completions and the 2026 pipeline were both set in motion during 2022. The completions decline from 608,000 toward 484,000 and below was locked in before rents softened, and any market tightening that follows the thinning pipeline is equally mechanical. Builders cannot restart the clock quickly either: land, entitlement, and construction add years between a rent signal and a response.
Related stories: What Heavy New Supply Does to Rent Growth: The Absorption Lag Explained · Sun Belt Versus Midwest Rental Yields: A Price-to-Rent Comparison.
Where Is New Supply Landing Hardest?
Deliveries land hardest where they cluster in submarkets rather than spreading across metros. Austin is the standard case: concentrated deliveries against healthy demand still produced measurable rent declines because thousands of units competed for the same leasing pool in overlapping quarters. The same concentration dynamics appeared at smaller scale in Nashville, Charlotte, and Phoenix through 2024-2025, per the pattern in Census regional construction data.
The distinguishing variable is absorption depth, not demand growth. A metro adding jobs quickly can still see rents fall if deliveries in one submarket exceed the whole metro's quarterly absorption capacity. A slower-growth metro with deliveries spread thinly across neighborhoods can absorb the same unit count without visible rent damage.
What Does a Shrinking Pipeline Do to Rents?
Standard supply logic: fewer deliveries mean less standing vacancy, and less standing vacancy restores pricing power to existing landlords — with the absorption lag running roughly nine to eighteen months behind the delivery change. The 2026 window is the pivot: completions falling from 608,000 toward roughly 484,000 and continuing lower implies progressively thinner competition from new stock through 2026 and 2027, per the Census construction series.
Two qualifications keep the logic honest. Single-family rentals and condos can add supply outside the multifamily count. And demand is not fixed — household formation responds to employment and to the buy-versus-rent price gap, so thinner supply only translates into rent support where demand holds.
How Should a Small Investor Read the 2026 Map?
The practical screen uses three public numbers per metro: recent completions from the Census Bureau's new residential construction tables, units under construction, and the local vacancy trend from the Housing Vacancies survey's regional detail. Markets with falling completions, shrinking under-construction counts, and vacancy that has stopped rising are the ones where the supply headwind is mechanically ending; markets where all three are still climbing are still eating the wave.
An illustration using the national figures: the roughly 124,000-unit drop in annual completions between 2024 and 2025 equals about two months of national absorption at recent rates — supply arithmetic of that scale moves vacancy rates by tenths of a percentage point nationally, and by far more inside the specific metros where the decline is concentrated.
Timing matters as much as direction. Buying into a supply-heavy market after deliveries have already peaked means buying after the worst rent damage is visible in the data; buying before the peak means holding through it. The three-series screen exists to tell those two situations apart.
What Should Readers Watch Next?
Watch the Census Bureau's monthly new residential construction releases for the completions and under-construction series, which lead the next two years of supply. Watch whether the national rental vacancy rate peaks and turns down from its 7.1 percent Q3 2025 level. And watch permits, the earliest series, for whether builders respond to any rent firming with a new starts wave that would restart the cycle. The data describes the aggregate stock; it does not price any single property, and none of this is a rent forecast.