Nobody Agrees on San Francisco Rent
What rent actually costs in SF, and why everyone is upset about it
Caroline Roche toured 25 apartments in a single week.
The 25-year-old demand planner had just taken a job in San Francisco and needed a one-bedroom she could afford. She wasn't picky about the neighborhood. She just needed a place to live. Twenty-five tours, back to back, in one of the tightest rental markets in the country.
Her experience, reported by The Real Deal, is not unusual in 2026 San Francisco. But what makes it worse is that before she even started looking, she had almost no way to know what apartments actually cost. The prices she found online told a very different story than what tenants actually pay.
Everyone reports a different number
Ask "what does a one-bedroom cost in San Francisco?" and you'll get a different answer depending on who you ask.
Median 1-bedroom rent in SF — same city, four answers
The spread is enormous: over $1,200 between the highest and lowest figure. None of these numbers are wrong, exactly. They just measure different things. Zumper tracks what landlords are asking for vacant units right now. The Census tracks what people are actually paying, including the retired teacher in the Sunset who's been in her rent-controlled apartment since 2004 and pays $1,400.
The problem isn't that the data is bad. It's that listing prices (what renters see) systematically overstate what apartments actually cost. Actual lease prices, concession terms, and renewal rates aren't published anywhere. Without that data, it's nearly impossible to know whether a price is fair.
What 143,000 SF households actually pay
The U.S. Census Bureau's American Community Survey doesn't ask what apartments are listed for. It asks what people pay. Across 15 San Francisco zip codes, covering 144,300 renter households, the distribution looks nothing like a single "average" would suggest.
Monthly rent paid by SF renter households
One in four SF renters pays $3,500 or more per month. But 14% pay under $1,000, a number that reflects rent-controlled apartments, subsidized housing, and SRO hotels that have been part of the city's fabric for decades.
The gap between those two groups is the story of San Francisco housing. And right now, that gap is getting wider.
The AI gold rush
San Francisco's rents are rising faster than any other major U.S. city. As of late 2025, one-bedroom rents were up over 15% year-over-year, and two-bedrooms were up over 20%, with medians rivaling New York City. The driver is no secret: artificial intelligence.
AI companies have been the dominant force in San Francisco office leasing, with the technology sector accounting for roughly a fifth of all office space leased in 2025. And the workers filling those jobs need places to live.
“My professional life is great. My living situation is, like, the total opposite.”
The competition has gotten fierce enough that some rentals now have bidding wars. The SF Standard reported on a 31-year-old tech worker who applied for a two-bedroom in Alamo Square listed at $5,000 a month. She and her roommate offered $5,100 and were told they'd made it to the "second round," then were asked to bid higher. They declined and lost the apartment.
The management company, Centron, called their process a "fairness process" for handling "overwhelming interest." The tenant who lost out called it something else: "It felt like it broke some unspoken covenant of renting."
Meanwhile, some AI startups have started leasing entire blocks of luxury apartments for their employees. Roy Lee, the 22-year-old CEO of Cluely, rented eight apartments in a single luxury high-rise, at $3,000 to $12,000 per month each, and gives them to employees for free. Flo Crivello, CEO of Lindy, offers about 40 employees a $1,000 monthly rent stipend if they live within a ten-minute walk of the office.
Brian Brown, a leasing agent with 20 years of experience in the city, told the SF Standard that inventory is the lowest he's ever seen. His brokerage's listings dropped below 15, a first in its history. Relocation consultants are making offers $2,000 over asking rent.
The supply drought
Demand is only half the story. The other half: San Francisco barely builds housing.
The city added roughly 1,700 new units in 2024, a twelve-year low. That's down from 2,500 in 2023, which was already down 30% from the year before. To hit its state-mandated housing target of 82,000 units by 2031, San Francisco would need to build about 13,000 units per year for the next six years. It's currently building at one-eighth that pace.
Over 20,000 approved units sit stalled in the pipeline. Developers say the math doesn't work. One told the SF Standard that gross revenue is "still around 5% below 2019, while expenses have climbed 30% in the past five years." Lenders find it cheaper to buy existing apartment buildings at a discount than fund new construction.
The consequences are predictable. When a city adds hundreds of thousands of jobs and barely any housing, rents rise. Not because landlords are uniquely greedy, but because too many people are competing for too few apartments. It's the simplest supply-and-demand story in economics, and one of the most painful to live through.
Why new construction helps everyone (eventually)
Housing economists have a term for what happens when new apartments get built: filtering. The idea is straightforward. When a higher-income renter moves into a new building, they leave their old apartment behind. Someone else moves into that apartment, leaving theirs behind. The chain continues.
It sounds abstract, but researchers have traced the actual chains. Evan Mast at the Upjohn Institute tracked 52,000 residents of new market-rate buildings across 12 U.S. cities using address history data. He found that for every 100 new market-rate apartments built, 45 to 70 people move out of below-median-income neighborhoods within five years. The chain reaction is real and measurable.
In San Francisco specifically, UC Berkeley researcher Kate Pennington found that within 100 meters of new construction, rents fall about 2%, displacement risk drops 17%, and landlords are 31% less likely to file eviction notices. Her study won the 2021 Urban Economics Association Prize.
The Sightline Institute puts it in simpler terms: the housing market is like a game of musical chairs. If there aren't enough chairs when the music stops, someone gets left out. You don't need to be fast. You just need to be rich enough to outbid the person next to you. Adding more chairs, any chairs, makes the game less brutal.
But here's the San Francisco problem: the city isn't adding chairs. At current construction rates, filtering barely has a chance to work. Without enough new supply to absorb high-income demand, the process runs in reverse: older buildings get renovated upmarket instead of becoming more affordable. As California YIMBY documented, a dearth of new housing means old houses are routinely upgraded into modern luxury rentals, eliminating the affordable stock that filtering is supposed to create.
The city's own economist, Ted Egan, projected that even full implementation of Mayor Lurie's Family Zoning Plan (which allows buildings up to 10 stories on west-side commercial corridors and up to 60 stories on Van Ness) would reduce monthly rents by only $75 to $125. The math requires building at a scale San Francisco has never achieved.
The data gap
The housing shortage makes rents high. But there's a less visible force at work: the rental market has a serious information asymmetry. Professional property managers use sophisticated data tools to set prices, tools that renters don't have access to.
In 2022, ProPublica reported that a company called RealPage was selling algorithmic pricing software used across nearly 20 million rental units, about a third of the U.S. market. The software ingested nonpublic data (actual lease prices, vacancy rates, concession terms) from across properties and used it to recommend daily pricing for every available unit.
Property managers followed the recommendations 80 to 90 percent of the time. One operator reported raising rents more than 25% within 11 months of adopting the software. Camden Property Trust saw turnover rise 15 percentage points after implementing the system, but revenue grew 7.4%, a net $10 million gain.
“We said there's way too much empathy going on here. This is one of the reasons we wanted to get pricing off-site.”
The Department of Justice sued RealPage in 2024, alleging the software facilitated illegal price coordination among competing landlords. Deputy Attorney General Lisa Monaco called it "a modern way to violate a century-old law through systematic coordination of rental housing prices." In January 2025, the DOJ also sued six of the nation's largest landlords, including Greystar.
A proposed settlement in November 2025 bars RealPage from using competitors' nonpublic data and requires a three-year court-appointed monitor. RealPage paid no fines and admitted no wrongdoing. San Francisco banned algorithmic rent-setting software outright in October 2024, the first city in the country to do so.
How concessions work (and why they're confusing)
Beyond pricing software, concessions (free months of rent) are a standard industry practice that renters often misunderstand.
When a building offers "two months free on a 14-month lease," the net effective rent is lower than the listed price. A $3,200 listing with two free months on 14 months works out to $2,743 per month effectively. But the lease is written at $3,200. When renewal comes, the renewal is based on $3,200, not $2,743.
During the pandemic, renters who got big concession discounts later received renewal increases of 30 to 60 percent, not because the "base rent" technically changed much, but because the concession vanished. A StreetEasy survey found that 40% of renters were confused by the term "net effective rent." That confusion matters, and it's worth understanding why concessions work this way. A building's valuation is based on its rent roll, the official price on each lease. A permanent rent reduction shrinks the rent roll, which lowers the property's appraised value and can violate bank loan covenants. So concessions serve a real financial purpose for building owners, but the result is that renters often don't realize what their actual long-term cost will be.
Lease timing
Property management publications discuss "lease expiration management," the practice of structuring lease terms so renewals fall during high-demand months. If too many leases would expire in February (low demand), new tenants get an 11- or 13-month lease to push their renewal into spring or summer. The industry describes this as "critical for NOI stability".
None of these practices are illegal. They're standard in professional property management. But together, they illustrate the information gap: the industry has pricing algorithms, market analytics, lease timing strategies, and comp data. Renters have listing prices and guesswork.
As the Urban Institute noted: "Rents are too high. Here are three ways to get the data we need to fix that." The fundamental problem is structural: home sale prices are recorded in public deed records, but lease prices are not. Without better data, renters can't make fully informed decisions.
Who gets squeezed
The combination of surging demand, stagnant supply, and a lack of pricing transparency has consequences that fall hardest on the people with the least margin.
Eviction notices for nonpayment of rent tripled in San Francisco between 2024 and 2025, from an average of 14 per month to 49 per month, according to the SF Rent Board. The hardest-hit neighborhoods: the Tenderloin, Northern SoMa, and the Mission, areas where the AI boom's office demand meets some of the city's most vulnerable populations.
“A lot of what might explain a rise in evictions is simply poor people getting poorer and becoming unable to pay their rent, and working-class people falling into poverty.”
Owner move-in evictions are up 39%. Breach-of-lease evictions are up 76%, according to the SF Rent Board Annual Eviction Report. Eviction filings across the city reached a decade high in 2025, with over 3,470 filings in a twelve-month period.
Mission Local reported on Gabriel, who pays $875 a month to live with four roommates in Crocker-Amazon. He earns about $1,300 from a part-time job. Rent takes 67% of his income. One woman arrived in San Francisco for a new job with an apartment lined up, but the landlord rented to someone else at a higher price. After five weeks of searching, she gave up on the city and moved to Daly City.
The rent-controlled housing stock (roughly 170,000 units built before 1979, representing about 70% of San Francisco's rental units) provides a crucial buffer. Annual increases are capped at 1.4% for 2025-2026. But that cap only protects you as long as you stay. When a rent-controlled tenant leaves, the landlord resets to market rate. The gap between what a long-term tenant pays and what a new lease costs has grown so wide that it creates enormous pressure (through buyout offers, eviction notices, and simple neglect) to push existing tenants out.
Where do displaced residents go? Migration data shows a steady flow from San Francisco to the East Bay. A net 14,800 people moved to Alameda and Contra Costa counties. From there, the chain continues outward: Alameda County residents move to Contra Costa, then to Solano and San Joaquin counties, each step further from the jobs and communities people left behind.
The neighborhood divide
The citywide average obscures how dramatically rent varies by neighborhood. Census data shows the percentage of renter households paying $3,500 or more per month, a rough proxy for market-rate in a newer or premium building.
Households paying $3,500+/month by SF neighborhood
Nearly half of all renters in Rincon Hill pay $3,500 or more. In the Tenderloin, it's 8%. These two neighborhoods are less than two miles apart. The Tenderloin's low figure reflects its concentration of SRO hotels, subsidized housing, and older rent-controlled buildings, exactly the housing stock that comes under pressure when market-rate rents in neighboring areas rise.
SoMa West (94103) tells a particularly striking story: 23% of renters pay $3,500+, but there's also a noticeable cluster of households paying $300-350, the footprint of the neighborhood's SRO hotels and legacy affordable housing, sitting alongside some of the city's most expensive new construction.
The transparency gap
Here's what all of this comes down to: in 2026, a renter searching for an apartment in San Francisco has almost no way to know what apartments actually lease for. Listing prices may include concessions that vanish at renewal. The "average rent" you Google could be off by $1,200 depending on the source. There is no public record of what any lease was signed for.
Salary transparency changed how people negotiate compensation. Glassdoor, PayScale, and Levels.fyi gave workers data to make better decisions. Home price transparency (through public deed records, Zillow, Redfin) gave buyers information to comparison shop with confidence.
Rent has no equivalent. And that makes it harder for everyone. Renters can't tell if a price is fair, and good landlords who price competitively can't differentiate themselves from those who don't.
RentSpy exists to close that gap. When renters share what they actually pay (the real lease price, the concessions, the fees, the renewal increase), it builds a dataset that helps everyone make better decisions. Not estimates. Not listing prices. What people actually signed for.
Better data makes the rental market work better, for renters trying to find a fair deal and for a market that runs more efficiently when pricing is transparent.
Data in this article comes from the U.S. Census Bureau American Community Survey (2022), Zumper, Apartment List, RentCafe, and original reporting by SF Standard, Mission Local, ProPublica, and The Real Deal. RentSpy collects actual lease prices from tenants. Submit yours to help build a more transparent rental market.