How the City Can Actually Build Enough Housing
Tobias Peter
Co-Director and Senior Fellow
American Enterprise Institute Housing Center
AEI.org/housing | May 2026
San Francisco needs roughly 10,000 homes per year—but is nowhere close
A shortage of ~100,000 homes
Needs to build ~10,000 housing units per year*
Very little family-sized, market-rate housing
Result:
Families are leaving (SF’s share of families living in it is half the metro average)
The working and middle classes are being pushed out
This is not a mystery. San Francisco doesn’t build housing because it makes it too hard to build.
* Per California’s current Regional Housing Needs Assessment (2023-2031)
The pipeline is not production
Pipeline category
Units
What it means
Under construction
~3,000
Actually being built
Permits issued / approved / filed
~11,400
Closer, but not guaranteed
Building permits not yet filed & applications under review
~25,900
Still exposed to delay, redesign, financing, and feasibility risk
Major multi-phased projects
~33,800
Long-horizon megaprojects
Total pipeline
~74,000
Not equivalent to near-term production
The city’s large housing pipeline is often cited as evidence that production will catch up over time. But the headline number can be misleading.
San Francisco reports roughly 74,000 homes in its pipeline:
Only ~3,000 are actually under construction.
The rest are spread across earlier permitting stages, projects still under review, and long-term multi-phased developments that may take years or decades to complete.
The overall pipeline has remained above 70,000 homes since 2020, but that has not translated into sufficient housing production.
Nearly half of the pipeline is in major multi-phased projects. These projects matter, but they are often infrastructure-heavy, capital-intensive, and slow to deliver.
Source: San Francisco Planning Department, Housing Development Pipeline Report, various years.
Bottom line:
San Francisco does not lack housing plans. It lacks a housing production system that lets ordinary projects pencil and get built at the scale needed.
San Francisco’s reforms are too narrow to solve the problem
Directionally right steps
Legalized fourplexes in single-family zones (up to 6 on corner lots) in 2022/3
Family Zoning Plan in 2025/6
Some low-density areas now allow up to 40 ft in height
Some additional height in certain corridors, but only comes with big strings attached (affordable-housing, labor, and design requirements)
Permitting appears to be getting faster.
Some of this reflects local process improvements, but some also reflects state streamlining laws. SB 35 pushed qualifying projects into ministerial approval, especially affordable housing. SB 423 could broaden that pathway to qualifying mixed-income and market-rate projects.
But the reforms remain limited
Too much of the city is still protected from meaningful change
Some added height comes with costly strings: affordability, labor, design, and process requirements
Faster permitting helps, but it does not guarantee projects will pencil or start construction. The application process seems excessive and tilts the playing field toward larger developers.
Bottom line:
Good steps, but insufficient scale. San Francisco cannot solve a citywide shortage with reforms that leave too much of the city off-limits or economically infeasible.
The Housing Abundance Success Sequence
Lessons distilled from over 30 case studies.
Allow enough density
Let the market produce at a scale that meets demand.
Make it by-right.
If a project complies with clear, objective standards, it gets approved—no discretionary reviews, no neighborhood veto, no delays.
Keep it short and simple (KISS).
Stop micromanaging design and process. Fewer steps. Clear limits. Lower costs. Fast decisions.
Bottom line:
Where cities follow this sequence, housing gets built. Where they don’t, it doesn’t.
San Francisco can’t grow while staying the same
San Francisco needs far more housing.
But much of the city still allows only modest change.
From 2012-2025, ~20 units/year have been demolished and rebuilt as 2 units on for every unit lost.
Keeping full local control means every neighborhood can say “not here” or “not much.”
The result is too little housing, too slowly, at too high a cost.
The city can’t do both:
Build enough housing and preserve low-density neighborhoods largely unchanged.
Bottom line:
San Francisco’s shortage is largely self-inflicted: the city has constrained where, how much, and how predictably housing can be built.
In one of the highest-land-price cities in the country, large areas of two-story homes cannot remain largely unchanged if San Francisco wants to solve its housing shortage.
The goal is not to eliminate neighborhoods, but to allow gentle, predictable redevelopment: 4–6 units, small apartment buildings, and family-sized infill where only limited change is feasible today.
Example: Low-density housing in the Sunset District
Source: Google and AEI Housing Center.
A realistic path to about 10,000 units per year
Assumptions:
Unlock already-legal multifamily by removing barriers: Assumes roughly doubling the rate of existing production.
Multifamily in commercial areas or at moderately higher density than currently allowed: 2,500 parcels with a median 8,000 sq. ft. lot * 2% conversion per year * avg. 40 units per lot.
Family housing infill: 2,000 sales of 2,500-5,000 sq. ft. lots / year * 25% infill conversion rate * avg. 5 units per lot.
Vacant land development: 1,300 buildable parcels (accounting for slope) * 10% development rate * avg. 5 units per lot.
Source of Housing Growth
Units per Year
Current production
~2,500
Unlock already-legal multifamily by removing barriers
+2,000
Multifamily in commercial areas or at moderately higher density than currently allowed
+2,000
Family housing infill
+2,500
Vacant land development
+650
Total potential
~9,650
Target (RHNA)
~10,000
San Francisco can reach its housing target, but not under today’s rules.
The real question is whether the system will allow enough housing to be built.