San Francisco Rents Are Up 22% This Year — Mesa, Arizona Is Down 7%. New Data Shows the 'National Rent Recovery' Is a Myth

Two Rent Reports Landed This Week. They Tell Completely Different Stories About the Same Market.

If you've been reading rent headlines lately, you've seen the same reassuring line over and over: "rents are stabilizing," "the multifamily correction is ending," "landlords are finally getting relief." That framing comes from national averages, and national averages are lying to you right now.

The new Zumper National Rent Report and the new Apartment List National Rent Report both dropped this week, and buried under the topline numbers is a story that matters a lot more than "rent is up 0.4% nationally": some metros are re-accelerating into double-digit annual rent growth while others are still in free fall. If you're underwriting a deal off a national average, you're either missing the best rent-growth markets in the country or walking straight into one where rents are still dropping.

This isn't a minor rounding difference. Zumper's data shows roughly a 29-point spread between the hottest and coldest one-bedroom markets they track this month. That's the difference between a deal that cash-flows immediately and one that needs two years of flat rent before you break even.

For most of the last three years, that gap didn't matter as much, because almost every market was moving the same direction: down, or flat at best, as the 2024 construction wave worked through absorption. What's new this month is that the markets have split. Some have absorbed their new supply and started repricing upward fast. Others built even more than their local job growth could support, and they're still cutting rent to fill units. Reading the national number today tells you nothing about which side of that split your deal sits on.

The Split: Supply-Starved Cities Are Repricing Fast, Oversupplied Ones Are Still Falling

Zumper CEO Shawn Mullahy put it plainly in the report: "The biggest story in housing is still supply... rent trends are diverging sharply by market, and the common denominator is inventory." Where new multifamily supply is scarce, landlords are pushing rent hard. Where builders overshot, rents are still being cut to fill units.

On the supply-starved side: San Francisco one-bedroom rents are up 21.9% year-over-year and two-bedrooms are up 22.6%, per Zumper — the fastest growth of any large market it tracks. Oakland is up 6.2%, San Jose up 2.6%. Apartment List's independently built index shows the same two cities moving the same direction — up 7.4% year-over-year for San Francisco and 6.1% for San Jose — even though its methodology produces different absolute numbers. Two data providers, two different models, the same conclusion: Bay Area rent growth is real and accelerating.

On the overbuilt side, Arizona is the standout warning sign. Zumper has Mesa down 7.3% year-over-year, Phoenix down 4.0%, Glendale down 1.9%, and Tucson down 1.1% — only Scottsdale (+5.9%) and Gilbert (+0.6%) are positive. Apartment List's numbers point the same direction across the broader Sun Belt: San Antonio leads national declines at -5.0% year-over-year, with Austin close behind at -4.3%. Both markets got hit by the same wave of new multifamily construction that, per Apartment List, peaked nationally in 2024 at more than 600,000 units delivered — the most in a single year since 1986.

The Numbers (What the Data Shows)

Here's the topline-versus-reality gap in one place:

• National average, Zumper: one-bedroom rent at $1,526, up just 0.4% year-over-year — the first positive annual reading since May 2025.
• National average, Apartment List: median rent at $1,385, actually down 1.2% year-over-year, and still 4% below its 2022 peak.
• Winners: San Francisco (+21.9% YoY 1BR per Zumper, +7.4% per Apartment List), Oakland (+6.2%), Scottsdale (+5.9%).
• Losers: Mesa (-7.3% YoY, Zumper), San Antonio (-5.0%, Apartment List), Austin (-4.3%, Apartment List), Phoenix (-4.0%, Zumper).
• Vacancy inflection: the national multifamily vacancy rate has ticked down to 7.2%, per Apartment List — the first decline since late 2021, after peaking at 7.3% in February. That's a real signal the historic construction wave is finally getting absorbed, but it's happening market by market, not all at once.
• Leasing speed: units are taking 30 days to lease on average, still three days slower than June 2025, per Apartment List — soft-market friction hasn't fully cleared even where rents are turning up.

Put it together and the "recovery" is a Bay Area and Midwest story — Milwaukee, Chicago, and Minneapolis are all posting steady positive growth per Apartment List — not a national one. The "correction" is a Sun Belt story, and it isn't over.

That split has direct underwriting consequences. A landlord in San Francisco or Oakland who renews a lease flat this year is leaving money on the table — comparable units are re-pricing 6-22% higher depending on the data source. A landlord in Mesa or San Antonio who underwrote next year's pro forma assuming last year's rent growth continues is about to get a very unpleasant renewal season. Same asset class, same country, opposite problems.

Common Mistakes Investors Make Here

• Underwriting rent growth off the national number. A 0.4% or -1.2% national average tells you almost nothing about what a specific zip code will do next year. Use metro-level data, not the headline.
• Assuming last cycle's "hot market" is still hot. Austin and Phoenix were the poster children for pandemic-era rent growth. They're now leading the national declines. Betting on their 2021 reputation in 2026 is working against you.
• Ignoring the supply pipeline. The 600,000-unit 2024 delivery wave, per Apartment List, is the entire reason Sun Belt rents cratered. If you don't check what's still under construction near your deal, you're underwriting blind.
• Treating vacancy and rent growth as the same signal. Vacancy is finally improving nationally, but that doesn't mean rents are rising everywhere at once — the two metrics are decoupled market by market this cycle.

How to Use PropGPT for This

"Pull the trailing 12-month rent growth and multifamily permit/delivery data for [metro or zip code], and tell me whether the current supply pipeline suggests rents there are more likely to follow the San Francisco pattern or the Phoenix pattern over the next 18 months." This forces PropGPT to connect supply data to a specific rent-growth trajectory instead of guessing off a national average.

"Compare rent growth and vacancy trend for [Metro A] vs. [Metro B] over the last 24 months, and flag which one is decelerating fastest." Run this before you commit capital to a market you haven't stress-tested against a comparable one.

"Given a property in [zip code] with current rent of $X, model three cash flow scenarios for next year: flat rent, -5% rent, and +5% rent. Show me the break-even point on my mortgage payment for each." Turns the market-divergence data into a number you can actually underwrite against, instead of a headline you read and forgot.

"Search recent multifamily construction permits and completions within 5 miles of [address], and estimate how many competing units will hit the market in the next 12 months." This is the local version of the "600,000 units delivered" story — you want to know if your submarket is about to get its own oversupply wave.

"Summarize this week's Zumper and Apartment List rent reports and tell me which three metros in [region I invest in] moved the most, in either direction, month over month." A standing prompt you can re-run every time a new national rent report drops, so you catch the divergence before it shows up in your own P&L.

The Bottom Line

The national rent number in this month's headlines is real, but it's an average of two opposite stories happening at once: supply-starved coastal metros re-accelerating into double-digit growth, and overbuilt Sun Belt metros still cutting rent to fill units built during the 2024 construction peak. If you're underwriting a deal — a new purchase or a renewal on one you already own — off the national number instead of your specific submarket's trajectory, you're one bad assumption away from a cash flow problem. Pull the metro-level data before you sign anything, and re-run it before every renewal cycle. The next 12 months will separate investors who noticed this split early from the ones still quoting last year's "hot market" list.