Map: DataSF · Analysis Neighborhoods

San Francisco already faces a housing crisis. A new wave of AI wealth could put even more pressure on the city.

Sources: Zillow [5], SF Planning [6]

Average rent 2025

Housing units 2025

Can building more market-rate housing solve this crisis?

Marina Safeway Project

Source: Rendering by Arquitectonica [7] · Quotes: [8]

A proposal in the Marina district to build 848 new market-rate homes.

However, the neighborhood isn’t happy about it.

“I don’t think it’s anti-housing. It’s anti-that-huge-project-that’s-really-terrible.”

“…it’s like this jarring monstrosity.”

“It’s not about views. It’s really about the fact that we have a weak infrastructure in the rain, and we notice it floods every time it rains.”

So, does building more market-rate housing actually lower rents in a meaningful way?

It depends.

Research shows new supply can bring rents down in three ways.

  1. Local effectsEffects in the nearby location
  2. Migration chainsPeople move to new homes, leaving vacant homes
  3. FilteringOlder buildings get cheaper over time

Each one works, but slowly and modestly [9]. And a flood of concentrated AI wealth can weaken all three.

Local effects

How does a new home affect its immediate neighborhood?

A new apartment building surrounded by older homes. A dashed circle marks 100 meters around it.
  1. −1.2–2.3%reduction in rents [10]
  2. −17%reduction in displacement risk [10]
  3. +29.5%increase in gentrification [10]
  4. +4.4%increase in rents in the cheapest nearby buildings [11]

Migration chains

Where do people go when a new home opens up?

A chain of moves: someone moves into the new unit from an older home, and someone else moves into that home from a lower-income area.
  1. Someone moves into the new unit, leaving an older home behind.
  2. Someone else moves into that one, and so on.
  3. For every 100 new units, 45–70 people move out of below-median-income neighborhoods, mostly within five years [12].

  4. One catch: this doesn’t work if the new tenant comes from outside SF.

Filtering

What happens to homes as they age?

  1. When it’s built
  2. 15 years later
  3. 30 years later
  1. As homes age, they tend to get cheaper, passing from higher- to middle- to lower-income households [13].
  2. It’s a slow process that takes decades [9].
  3. Economists call this downward filtering.

  4. During periods of high demand, the opposite can happen.

    Older homes filter up to higher-income households instead [13].

    This is upward filtering.

Final section (WIP)

My plan for this section is a choose-your-own scenario based on different user-selected policy choices. Still thinking about what those are.