For the better part of three years, I sat across from LPs, listened to colleagues at conferences, scrolled through breathless takes from analysts who had never swung a hammer or signed a personal guarantee — and heard the same thing on loop: San Francisco is over. The Bay Area is done. Anyone still investing there is delusional.
I kept buying. I kept building. And I kept my mouth mostly shut, because the data would eventually do the talking for me.

March just did the talking.

The Numbers

San Francisco posted the highest annual apartment rent growth in the United States at 6.8% — not top five, not top three — first. Average asking rents hit $3,460, nearly double the national average. Vacancy fell to 4.1%, the tightest it’s been in nearly a decade. Net absorption came in at 3,200 units over the past year. New supply? 1,500 units delivered. That gap — 1,700 units of unmet demand — is not a blip. That is a structural condition.

Mission Bay/China Basin and SoMa posted annual rent gains above 14%. Fourteen percent. In submarkets that every macro tourist declared “permanently impaired.” Luxury properties — four and five-star assets — are averaging above $4,400 per unit and hitting near-full occupancy shortly after certificate of occupancy. The market isn’t just recovering. It’s repricing upward.

What the Doom-Loop Crowd Got Wrong

They modeled a city. I understood an ecosystem.
San Francisco was never going to be replaced by Austin or Miami or Boise — not because those markets aren’t excellent on their own terms, but because they don’t have what the Bay Area has: the deepest concentration of AI talent, venture capital, and technology infrastructure on the planet. When the AI hiring wave began in earnest, those workers didn’t scatter to the Sun Belt. They moved to Mission Bay. They moved to SoMa. They moved to the neighborhoods the media had declared dead.

Return-to-office mandates accelerated the math. Companies like Salesforce, OpenAI, Anthropic, and dozens of well-funded Series B and C shops didn’t just hire — they mandated presence. That converted remote workers back into renters. It converted renters into demand. Demand into absorption. Absorption into pricing power.

High-income transplants from New York and Los Angeles — workers chasing the AI opportunity — didn’t come to rent a studio in the Tenderloin at a discount. They came to live in Mission Bay luxury product at whatever the market required. And the market required a lot.

Why I Never Wavered

I’ve been doing this for 25 years. I’ve seen the death narrative play out in New York after 9/11. After 2008. After COVID. Every time, the obituary writers were wrong and the operators who stayed long were right.

San Francisco is a harder case — I’ll give the skeptics that much. The political dysfunction is real. The permitting environment is genuinely broken. The cost of delivery is the highest in the country by most measures. These are not invented problems.

But here’s what the models missed: supply destruction is a feature, not a bug, when you’re long the market. Every project that got killed in entitlements, every developer who pulled out, every lender who went risk-off on Bay Area multifamily — they were doing my underwriting for me. They were tightening the vacancy that is now at a ten-year low. They were building the rent growth I’m capturing today.

Broken supply pipelines in high-demand markets are not a reason to exit. They are the investment thesis.

Where I am focused now

The submarkets with the most runway are the ones that have the combined tailwinds of improving safety perception, proximity to AI/tech campuses, and severe supply constraints. Downtown and Civic Center — not the obvious answer, but watch those numbers. Mission Bay is already repriced. The next wave of rent growth will be captured in the submarkets that still carry a discount to fundamentals but are trending hard in the right direction.

Luxury new construction is not a crowded trade when nobody has been building it. The four and five-star product posting 9.6% annual gains didn’t appear by accident. It appeared because most developers walked away. The ones who didn’t are collecting that premium now.
I’m not interested in relitigating the last three years. The data is doing that on its own.

What I am interested in is the next three — and what the current supply pipeline tells me about where rents are heading when the 2026 and 2027 deliveries come in well below what demand requires.

The narrative will eventually catch up to the fundamentals. It always does. By then, the trade will be much more expensive.

— Daniel Kaufman
The Kaufman Report