1.2 million lease-ups are the ceiling on apartment rents

The apartment market has been waiting for the supply wave to pass, and the wave has visibly crested: at the start of August, 1.2 million units were in lease-up nationwide, down from the early-2025 peak of 1.4 million. The trouble is what sits between a peak and a normal market: that 1.2 million is still about double the average of the previous decade, the plainest available explanation for why operators' ability to raise rents remains suppressed while demand holds up.

Paul Fiorilla, director of research at Yardi, put numbers to that mechanism in a Sept. 4 analysis: rent growth is "highly correlated" with the number and share of new building stock in lease-up in a given market, which makes the contest one between landlords rather than between landlords and renters. "Even though demand is strong in most of these markets, just the sheer number of new units that need to get filled is really high," Fiorilla told Multifamily Dive. "So it's taking longer to do it." Renters with options, he added, mean "you can't charge as much as you otherwise would."

Jay Lybik, senior director of market research at Continental Properties, put the scale in context: "this is the most units we've had delivered in multifamily since the mid 1980s," he told Multifamily Dive, while cautioning that conditions then are not directly comparable to today's. What he does compare is expectations to outcomes: "there are so many properties that are still in lease-up, and the impact of that has been much stronger and longer than anyone anticipated in the market."

That word, anyone, covers a good deal of underwriting. When the industry began digesting the pipeline, the reasonable assumption was that lease-up pressure would track deliveries down, and deliveries are indeed falling; the number of properties competing for the same renter this autumn has not fallen with it.

Discounts are doing the work that rent growth would otherwise do: concessions remain higher than usual, according to both Lybik and Fiorilla, and they are no longer confined to new buildings. Lybik pointed to recent RealPage data showing about 25% of units in buildings constructed in the 1990s offering concessions, up from 18% three years ago—a share that sits well below what 2020-vintage properties are handing out, he told Multifamily Dive, but higher than it was.

Fiorilla's account of why that persists is the sharpest part of the analysis: operators, he says, are frustrated because "they need to offer concessions in order to retain — not only to attract, but also to retain — tenants." The discount has migrated from an acquisition cost to a retention cost, and retention costs do not evaporate when a market tightens. "As the market improves, concessions will diminish, but it's going to go down slowly because people have gotten used to it. They expect them, and if they don't get them, they'll look somewhere else."

That is the difference between a soft market, which returns to the old rent roll once supply is absorbed, and a repriced one, in which the climb back happens one renewal at a time against a renter who now reads the discount as the asking rent and its removal as an increase. Lybik's description of operator behavior, "trying to fill the units any way they can," is what a market looks like when the tenant sets the effective rate.

The operating story becomes the capital story here: as this publication has argued, apartment capital is clearing at public data points now, and buyers are underwriting operations rather than rent growth. The lease-up count is the data point that enforces that discipline: a sponsor modeling market rent growth as an input is modeling the one variable it cannot manage, while renewal retention, concession structure and expense control are levers a manager can pull.

Our August reporting caught the turn already: for the second straight month fewer apartments were carrying concessions, and first-half construction starts hit their lowest level since 2012. Both facts point the right way, but neither is on the right clock: starts are a revenue line for the end of the decade rather than for this leasing season, and the supply that actually competes with a 1990s-vintage building for a tenant right now is the 1.2 million units in lease-up. The overhang has shrunk by roughly 200,000 units since early 2025, and simple extrapolation of that pace puts normalization years out, though extrapolation is a rough tool when deliveries are also falling.

Fiorilla and Lybik both see conditions improving, and neither puts a date on it. The series that will supply the date is the one Lybik cites: when concessions in 1990s-vintage buildings fall back toward the 18% registered three years ago, the sector will have absorbed enough new stock for a landlord to price the building rather than the competition down the street. Until then, the concession is the rent.