Skip to main content

San Francisco's AI Rent Emergency: What It Means for Multifamily Investors

By Avi Hacker, J.D. · 2026-09-12

What is San Francisco's AI rent emergency? It is the designation Mayor Daniel Lurie announced on September 10, 2026, alongside tenant protection measures responding to rent increases and eviction filings the city attributes to the artificial intelligence hiring boom. For multifamily owners, the "10% rent cap" circulating in national coverage is far narrower than it sounds, and that distinction decides whether a San Francisco value-add business plan still pencils. For the underlying framework, see our complete guide to AI multifamily underwriting.

Key Takeaways

  • San Francisco declared a rent emergency on September 10, 2026, citing a 25.6% rise in median rents and more than 1,000 eviction notices filed this year.
  • The widely reported 10% cap applies only to banked increases and capital improvement passthroughs on rent-controlled units, not to market rents citywide.
  • Capital improvement passthroughs are the exact mechanism value-add sponsors use to recover renovation capital on older San Francisco stock.
  • CBRE reported Bay Area multifamily rent growth of 7.7% year over year in Q2 2026, far below the 25.6% median asking figure the city cites.
  • Newer construction sits outside both San Francisco rent control and the state's 15 year exemption, concentrating the impact on pre-1979 buildings.

What San Francisco Actually Declared

The declaration itself is symbolic and grants no authority to set rents. What matters to investors is the legislative package the Office of the Mayor released alongside it.

The measures fall into two groups. On eviction, Supervisor Jackie Fielder's legislation would block nonpayment evictions unless a tenant owes at least one month of federally set fair market rent, the city added $3 million this year to its Tenant Right to Counsel program to move roughly 400 households from partial to full scope representation, and Ellis Act displacement payments rise 25%, or nearly $3,000 per person.

On rent, Supervisor Danny Sauter's measure caps annual allowable increases from banked increases and capital improvement charges at 10% for rent-controlled tenants, effective upon introduction. Supervisor Matt Dorsey's measure requires annual tenant notification of increase limits and just cause rules, and City Attorney David Chiu is directing renters to a complaint portal for systemic landlord violations.

Why the 10% Cap Is Narrower Than the Headline

Sauter's cap does not touch market rents and does not apply to every building. It limits two recovery mechanisms available to owners of rent-controlled units: banked increases, which are unused allowable increases carried forward from prior years, and capital improvement passthroughs, which recover part of qualifying renovation spending through tenant rent.

That narrowness is not good news for value-add sponsors. It is the opposite. Capital improvement passthroughs are precisely the lever a value-add business plan on older San Francisco stock depends on, because rent-controlled units cannot simply be marked to market on renewal. A sponsor who underwrote renovation capital assuming multi-year passthrough recovery plus banked increases now faces a hard 10% annual ceiling on that combined recovery. Meanwhile, newer construction is largely unaffected: San Francisco's rent control ordinance generally reaches buildings with certificates of occupancy before June 1979, and California's statewide rent regulation carries a rolling 15 year exemption for new construction that the city is now requiring landlords to disclose.

The result is a widening spread between two kinds of San Francisco assets. Post-1979 product keeps its pricing power into an extraordinarily tight market. Pre-1979 value-add stock keeps the same renovation cost basis while losing part of its recovery runway, pressuring stabilized NOI and, at any given cap rate, the exit value the model assumed.

The Two Rent Numbers, and Why They Disagree

The city cites a 25.6% increase in median rent over the past year. CBRE's Bay Area Multifamily Figures for Q2 2026 reports rent growth of 7.7% year over year across the Bay Area, with the San Francisco and Peninsula submarket leading at 11.4%. Both can be accurate because they measure different things. Median asking rent tracks what newly listed units advertise, which reprices instantly when high income AI hires compete for a thin pool of available apartments. Institutional rent growth series track effective rents across stabilized, largely occupied stock, where rent-controlled units and existing leases damp the swing.

For underwriting, the institutional series is the better input for in-place revenue and the asking rent series the better read on marginal demand and loss to lease. Using 25.6% as a rent growth assumption would badly overstate achievable revenue, particularly now that the passthrough ceiling limits how fast rent-controlled units can converge toward market.

The supply picture explains why the city acted. CBRE reported Bay Area vacancy compressing to 2.8% in Q2 2026, San Francisco at 2.4%, and 884 units delivered against 5,104 units of net absorption, roughly 5.8 to 1. The mayor's office cited citywide multifamily vacancy of 2.2% as of June 2026, against Oakland at 4.5% and Chicago at 6.5%, and noted that 2024 housing authorizations were the lowest since 2010.

What Changes in Your San Francisco Underwriting

Four adjustments follow directly from the package:

  • Re-run passthrough recovery schedules. Any model recovering renovation capital through capital improvement charges on rent-controlled units needs the combined banked plus passthrough increase capped at 10% annually. Longer recovery lowers near term NOI and compresses cash-on-cash return.
  • Re-price delinquency workouts. The Fielder threshold means nonpayment cases cannot proceed until arrears exceed a month of fair market rent, lengthening the bad debt tail and pushing collections loss higher.
  • Reserve for Ellis Act exits. Displacement payments rose 25%, adding nearly $3,000 per person, which matters to any plan contemplating withdrawal from the rental market.
  • Segment the portfolio by regulatory vintage. Pre-1979 and post-1979 assets now behave as separate risk classes. Blended market assumptions will mislead on both.

None of this makes San Francisco uninvestable. A 2.2% vacancy market with the largest AI employment base in North America has real fundamental strength behind it. But the regulatory overlay, not the demand story, is now the binding constraint on returns for one identifiable slice of the stock. CRE investors looking for hands-on AI implementation support can reach out to Avi Hacker, J.D. at The AI Consulting Network.

Using AI to Model Regulatory Exposure

This is a tractable modeling problem: classify every unit in a rent roll by regulatory status, then apply the correct increase ceiling to each.

A workable approach is to extract the rent roll and unit level construction dates, have a model such as Claude, ChatGPT or Gemini tag each unit as rent controlled or exempt based on certificate of occupancy date, then compute the achievable increase path under both the prior passthrough rules and the new 10% ceiling. The delta between those paths, carried to stabilized NOI and capitalized at your exit cap rate, is the dollar cost of the ordinance to the asset. Our walkthrough on AI rent growth projection covers the forecasting mechanics, our guide to renovation sequencing and capex prioritization is the piece a passthrough ceiling hits hardest, and our approach to rent control exposure analysis offers a parallel framework in another regulated asset class.

One caution worth stating plainly: a model will confidently assign a regulatory status it cannot verify. Certificate of occupancy dates and prior passthrough petitions must come from source documents, not model assumptions. Treat the output as a structured first pass a human confirms against the Rent Board record. If you are ready to build this regulatory screen into your underwriting, The AI Consulting Network specializes in exactly this.

Frequently Asked Questions

Q: Does San Francisco's rent emergency cap market rents at 10%?

A: No. The 10% cap in Supervisor Danny Sauter's measure applies only to annual increases derived from banked increases and capital improvement charges on rent-controlled units. Market rents on units not subject to rent control are unaffected by it.

Q: Which San Francisco buildings are affected?

A: The cap reaches rent-controlled units, which under San Francisco's ordinance generally means buildings with certificates of occupancy issued before June 1979. Newer construction also benefits from a rolling 15 year exemption under California's statewide rent regulation, which the Dorsey measure now requires landlords to disclose to tenants.

Q: Why do reported San Francisco rent increases range from 7.7% to 25.6%?

A: They measure different populations. The 25.6% figure the city cites tracks median asking rent on newly listed units, which reprices immediately. CBRE's 7.7% Bay Area figure for Q2 2026 tracks effective rents across stabilized occupied stock, where existing leases and rent control slow the adjustment.

Q: Is the emergency declaration legally binding on landlords?

A: The declaration itself creates no new landlord obligations. The binding changes come from the individual ordinances, which move through the Board of Supervisors, though Sauter's 10% cap was described as effective upon introduction.

Q: Should investors avoid San Francisco multifamily now?

A: Not on these facts alone. San Francisco reported roughly 2.2% multifamily vacancy in June 2026 against a Bay Area AI employment base CBRE puts near 98,700, the largest in North America. The package narrows returns on pre-1979 value-add strategies rather than impairing the market broadly. For guidance on underwriting regulated multifamily, connect with The AI Consulting Network.