San Francisco's Doom Loop Was Always a Myth

How the city went from doom loop to AI boom while everyone was writing its obituary

When I graduated college in 2012, San Francisco was barely a tech city. Tech made up only 10-15% of the city’s economy. I moved to SF in 2007 and worked my way through school as a busboy and security guard while the city went through hell. By 2009-2010, home values had dropped 45% from their peak, unemployment hit 10%, and homicides were running around 100 per year. Downtown office vacancy rates were 15%. Foreclosures were everywhere.

The finance industry had just imploded. The dot-com crash was only eight years behind us. Most people with money were staying away from tech. Salesforce and Twitter already existed, but that was about it for SF tech companies. Uber, Airbnb, Stripe, Coinbase, DoorDash, Pinterest—none of these existed yet. They’d all start in the next few years, but in 2008-2009, nobody knew they were coming. Y Combinator funded fewer than 25 companies in 2008. Today they fund over 500 per year.

By 2023, San Francisco was in crisis again. Covid had proven that tech workers didn’t need to live near their office. They left for Austin, Miami, Seattle, and cities across the Bay Area. After 2022, New York became the other big winner—pulling founders and investors who wanted to be in a city that felt alive again. The “Doom Loop” narrative peaked in 2023-2024. Five problems were breaking the city:

Cost of living. San Francisco became unaffordable. Tech workers looked at their paychecks and realized they were spending 40-50% on rent for a one-bedroom apartment.This didn’t happen overnight. In the early 1970s, American cities started restricting housing construction through zoning laws. Neighbors sued to block new buildings. In San Francisco, this went on for 50 years. Supply stayed flat while demand exploded.The tide started turning around 2022-2023. Pro-housing advocates began winning over the old guard who’d blocked construction for decades. But it’ll take until 2035 or so before enough new housing gets built to actually lower prices.

Office vacancies and small businesses. COVID didn’t just empty San Francisco temporarily. It broke something permanent. Office vacancy hit 35% in 2023, more than double the 2008 peak. But unlike 2008, workers weren’t coming back. Nearly half of San Francisco workers went remote during COVID. By 2023, 30% of SF job postings were still remote versus 12% nationally. The office workers just weren’t downtown anymore. Small businesses collapsed. By late 2020, 5,000 businesses had closed, 2,000 of them being permanent. San Francisco lost 6.5% of all its businesses by the end of 2021, the worst of any major American city. This wasn’t a recession. It was a structural break. The 2008 crisis was cyclical—people came back. This time, remote work meant they didn’t have to.

Homelessness. San Francisco spends $750 million per year on homelessness. That’s $94,000 per homeless person. Where does it go? Nonprofits running shelters and programs. Bureaucrats administering the system. Supportive housing that costs more than market-rate apartments. Meanwhile, thousands of people still live on the streets, many with severe mental illness or addiction who refuse help. The system has become self-perpetuating. More spending hasn’t reduced homelessness, it’s created an industry dependent on the problem continuing.

Fentanyl. San Francisco sees roughly 600 overdose deaths per year now. Fentanyl killed 451 people in 2024 alone. The crisis started around 2015 and accelerated through COVID. Fentanyl is synthetic, potent, and easy to smuggle. It replaced heroin almost completely. Open-air drug markets run in the Tenderloin and SoMa. People use it on the sidewalks in broad daylight. This overlaps with homelessness but it’s a separate problem. Many fentanyl users have housing. Many homeless people don’t use it. But the visible street crisis, the thing that makes San Francisco feel broken, is both problems happening in the same neighborhoods at the same time.

Crime. Two California policies collided to create a revolving door for repeat offenders. Prop 47 (2014) made theft under $950 and drug possession misdemeanors instead of felonies. In re Humphrey (2021) made it unconstitutional to hold people in jail pretrial if they couldn’t afford bail. Both policies solved real problems: prison overcrowding and wealth-based detention. But together they meant repeat offenders faced no consequences. Get caught shoplifting, walk out the same day, do it again tomorrow. Drug dealers work openly in the Tenderloin. Stores lock up basics like toothpaste. Walgreens closed 17 SF locations. The visible disorder eroded trust in the system. It’s a statewide problem, but density and open drug markets make San Francisco the worst example.

So San Francisco is broken, right? No. Despite everything I just described, the city is stronger than it’s been in the 18 years I’ve lived here. Here’s why.

Crime. Here’s the paradox: overall crime is at historic lows. Homicides hit a 60-year low in 2024: just 23 murders, down 55% from 2023. Violent crime dropped 14% in 2024 versus 6% in comparable cities. Car break-ins fell 41% in early 2025. San Francisco is safer than it was in 2008 when I arrived. It’s much safer than the 1990s. So what’s the problem? The repeat offender issue creates visible disorder—shoplifting, open-air drug dealing, property crime—that makes the city feel unsafe even though violent crime is down. A small number of prolific offenders cycle through the system and create an outsized perception problem.

AI as an economic engine. Remember when tech was only 10-15% of San Francisco’s economy in 2008? AI alone is now reshaping the entire city. AI companies occupy 5 million square feet of office space. OpenAI has nearly 1 million square feet in Mission Bay. Anthropic has over 300,000. The Bay Area captures 78% of all AI venture funding in the US, most of it going to San Francisco companies. If current projections hold, AI will add 50,000 workers by 2030 and quadruple its office footprint. That would cut the city’s office vacancy rate in half within five years. Y Combinator funds about 500 companies per year now. Roughly 85% choose to stay in San Francisco, even though most founders are immigrants who could work anywhere.

Tourism recovery. During the doom loop panic of 2023, people said tourists would never come back. They were wrong. Hotel occupancy reached 65% in 2025, up from 63% in 2024. The Moscone Center, empty during COVID, hosted 34 major conferences in 2025, a 64% jump that generated 657,000 hotel room nights. Domestic tourism has recovered. International visitors are returning. Next up: the 2026 Super Bowl and World Cup games in the Bay Area. San Francisco will be packed.

Mayor Lurie’s momentum. Daniel Lurie became mayor in January 2025, a political outsider who promised competence over ideology. So far, he’s popular: 73% approval rating. He’s raised $100 million in private funding. He’s launched a housing permitting tracker, a new homeless strategy, and pushed back against Trump’s threatened federal takeover of the city. It’s early. Crime was already dropping before he took office. His homeless plan hasn’t shown results yet. But the city feels different—like someone’s trying to fix things instead of managing decline.

San Francisco keeps winning the same way: it concentrates talent and capital in a way nowhere else can. The 1970s were rough. The dot-com crash wiped out fortunes. The 2008 crisis emptied offices. COVID looked terminal. Yet here we are again, with AI companies fighting for office space and founders choosing SF despite being able to build from anywhere. Why? Because SF offers something other cities don’t: density of the right people at the right time. You can’t replicate the concentration of AI researchers, experienced founders, and risk-tolerant capital. You can’t Zoom your way into the coffee shop conversation that becomes your co-founder or your first hire. The city has serious problems. But it has the one thing that matters for technology: it’s where the people building the future want to be.