A San Francisco couple earning a combined $365,000 a year just gave up trying to rent a one-bedroom apartment. That single detail, reported by The New York Times this week, is the cleanest signal yet of what an AI liquidity event does to the city that hosts it. Katrine Razniak, a 27-year-old who more than doubled her pay to $180,000 after joining Rippling, and her partner Adam Woodbury, a software engineer making $185,000, looked at roughly 30 listings over three months and lost every one. At a $5,200-a-month unit, 30 names hit the open-house sign-up sheet within an hour. They stopped looking. Woodbury moved to Lake Tahoe. They are now thinking about Seattle.
The human story is familiar. The business story underneath it is the one operators should care about: a small cluster of pre-IPO AI companies is repricing an entire local economy before most of them have even listed.
The numbers behind the squeeze
San Francisco’s overall cost of living now runs 65.6 percent above the national average, according to the Council for Community and Economic Research. Housing is the part doing the damage. The city’s median home price topped $1.7 million in April, against a national median near $450,000, and average apartment rent has passed New York to become the most expensive in the country.
The trend lines make the cause hard to argue with. National rents are essentially flat to falling, yet San Francisco rents are up roughly 18 percent year over year, the steepest in the nation, with brokers tying the spike directly to tech and AI demand. In the city’s most sought-after neighborhoods, the apartment vacancy rate has collapsed to about 3 percent from roughly 13 percent in 2020, while new construction has stalled. That combination, a sharp demand shock landing on a frozen supply, is why a top-20-percent household income now buys a roommate situation in Haight-Ashbury.
The category breakdown shows just how lopsided the pain is.
This is a liquidity story, not just a salary story
Here is the part that separates this boom from the last one. San Francisco’s chief economist Ted Egan framed it precisely: when Uber went public in 2019, the city’s marquee IPO was worth about $82 billion. The companies headquartered there now are an order of magnitude larger. Sacra, a private-markets research firm, estimates that OpenAI, Anthropic, and the newly public SpaceX could mint more than 20 new billionaires among current and former staff.
The wealth event is no longer hypothetical. SpaceX already priced its IPO in June 2026 at roughly $1.75 trillion, the largest in history, then saw the stock run up and give back about a third of its gains within two weeks. Anthropic filed confidentially in early June, raised $65 billion at a $965 billion valuation, and is targeting an October listing that could make it the first company ever to debut public at a $1 trillion cap. OpenAI filed within days but, per Bloomberg’s late-June reporting, is now leaning toward 2027 after watching SpaceX’s debut cool.
Stack those bars next to Uber and Egan’s point becomes visual. The last wave that reshaped San Francisco rents was a fraction of what is queued up now. Three listings could pull north of $200 billion out of public markets inside a single window; the entire U.S. IPO market raised about $45 billion in all of 2025. A meaningful slice of those proceeds converts to local purchasing power, and it lands on the same few thousand square blocks.
What actually drives the rent, and when
The instructive nuance for operators is the timing. Pre-IPO paper wealth does not move rent. Liquidity does, and so does the recruiting comp inflation that runs ahead of it. Mean annual pay in San Francisco hit $196,365 last year, up from $153,359 in 2020, because AI labs that do not need to be profitable can outbid everyone else for the same engineers. That bidding war resets the comp benchmark for every other employer in the metro, which is why a $185,000 software engineer can credibly feel priced out of his own market. The squeeze arrives through the labor market first, then gets amplified when the stock vests.
That is the mechanism worth internalizing. The same competitive dynamic that has these labs slashing token prices against each other and racing to absorb their own data-center power costs also plays out in talent and real estate. Uneconomic competition for inputs does not stay contained to the income statement. It spills into the cities these companies occupy.
The honest counterpoint
It would be lazy to pin all of this on AI. San Francisco has been one of the least affordable cities in America for two decades, and the proximate reason rent is climbing while the rest of the country sees rents fall is supply, not just demand. Roughly 3 percent vacancy and stalled construction mean any demand shock gets converted straight into price. AI wealth is the trigger. A housing market that cannot add units is the reason the trigger does so much damage. Cut the demand shock in half and a functional supply market would absorb it; this one cannot.
There is also a real question of whether the squeeze sticks. SpaceX’s stock already retraced 32 percent from its peak, OpenAI is delaying, and the broader read on the AI market, captured in the Stanford 2026 AI Index, is that the value these labs create is leaking to users and deployers faster than the labs capture it. If the listings underwhelm or the AI trade cools, some of this paper wealth never fully converts, and the rent curve could flatten as fast as it spiked.
Frequently Asked Questions
Why is San Francisco so expensive in 2026?
The city’s overall cost of living sits about 65.6 percent above the U.S. average, and housing is the main driver. A demand shock from AI hiring has landed on a frozen supply market, with prime-neighborhood vacancy near 3 percent and new construction stalled. That combination converts demand straight into price, which is why San Francisco rents are climbing while national rents are flat to falling.
How much do you need to earn to live comfortably in San Francisco now?
There is no clean threshold, and that is the point of the story. A household earning $365,000 between two people still could not secure a one-bedroom under $5,000 a month. Mean annual pay in the city reached $196,365 last year, yet workers in the top 20 percent of U.S. households increasingly describe a low-level financial vigilance rather than comfort. The honest answer is that “comfortable” depends heavily on whether you own, rent, or carry equity in an AI company.
Are OpenAI and Anthropic going public?
Both filed confidentially with the SEC in early June 2026. Anthropic is targeting an October 2026 listing that could make it the first company to debut public at a $1 trillion valuation. OpenAI filed within days but is now leaning toward 2027, according to late-June reporting, after SpaceX’s record IPO cooled. SpaceX itself already priced in June 2026 at roughly $1.75 trillion.
How much higher is San Francisco rent than the rest of the country?
San Francisco’s median rent now runs well over 150 percent above the national median, the highest of any major U.S. city, having recently passed New York. Average apartment rent in the city is up roughly 18 percent year over year, even as rents nationally are flat to slightly down.
Will San Francisco housing prices keep rising?
That depends on whether the AI wealth converts and stays. The bull case is that 20-plus new billionaires and a wave of newly liquid employees keep bidding up a fixed housing stock. The bear case is that SpaceX’s stock already retraced about 32 percent from its peak, OpenAI is delaying, and AI-lab economics are under pressure, any of which could slow the conversion of paper wealth into local spending and flatten the rent curve.
The Business Model Analyst Take
The headline writes itself as a story about whiny six-figure earners, and that framing misses the lesson. What San Francisco is showing in real time is how concentrated, pre-liquidity wealth distorts a local economy, and the distortion runs through the labor market before a single share trades.
For founders and operators, three things are worth pulling out. First, your cost of talent in an AI hub is now being set by companies that do not need to make money, which means location is a margin decision, not just a culture one. Second, liquidity events are local economic shocks with predictable second-order effects, and the smart move is to read the comp inflation ahead of the IPO rather than after. Third, watch the supply side wherever you operate, because a demand shock only becomes a crisis when supply is frozen, and that is as true for housing as it is for compute, talent, or any other input you depend on. The couple leaving for Seattle are not the story. The repricing mechanism they walked into is.
