Billboards advertising artificial intelligence companies seem to be everywhere in San Francisco, even as the threat of an AI bubble persists.
Justin Sullivan/Getty Images
Billboards advertising artificial intelligence companies are abundant in San Francisco, the epicenter of the technology. There are new concerns that the AI boom could become an economic bubble, but how would the city handle a burst bubble?
Justin Sullivan/Getty Images
San Francisco is flooded with abstract billboards from obscure artificial intelligence firms touting the merits of replacing humans while branded bus stops double as marketing launchpads for arcane startups.
It’s one of several signs that the city has become the global epicenter of the AI revolution, where the future is being shown in plain sight, but developed behind layers of secrecy. The surge in AI activity has given San Francisco’s embattled business reputation a much-needed boost, drawing fresh investment, talent and hype to a city still reeling from tech’s pandemic-era retreat.
As money pours into growing companies with murky missions and sky-high valuations, new office leases are being inked, a mansion frenzy is under way and developers are dusting off plans for major office skyscrapers. The AI enthusiasm will be in full force in San Francisco this week as Dreamforce, the city’s biggest conference, returns with more than 45,000 attendees and a projected $130 million in local spending.
Article continues below this ad
Yet, despite the excitement, there are clear signs this boom could actually be an economic bubble that echoes previous cycles, and analysts are warning that it could burst. But if that happens, don’t expect the city’s billboards to begin advertising soft drinks.
The city has endured three major crashes in the past three decades and each time emerged changed — yet still as home to the next big thing in tech. A new wave of innovation and investment always seems to follow a correction, remaking the city’s economy and skyline while renewing battles over housing costs and affordability.
“I think you could make an argument that it’s all super-inflated and it has to unwind itself,” Ted Egan, San Francisco’s chief economist, said of the AI market. “That doesn’t mean it’s going to be another dot-com crash and 10-year winter for Bay Area tech.”
In some ways, the question isn’t whether this latest boom goes bust, but how San Francisco will handle what comes next, and what the city will become because of it.
“The work for all of us is to make sure that the recovery that we’re driving in the city is durable and that we can create the conditions for the companies, for the workers, for the residents — and the people who work in other parts of the economy — to thrive,” said Ned Segal, the city’s chief of housing and economic development.
Article continues below this ad
Analysts believe San Francisco, epicenter of the artificial intelligence boom, could survive a bubble.
Justin Sullivan/Getty Images
Still, the worries over a bubble are real. The Bank of England and International Monetary Fund both warned recently of a market correction fueled by AI overenthusiasm. Deutsche Bank said last month that the splurge in AI spending is propping up the overall U.S. economy, but it isn’t sustainable. Researchers at the Massachusetts Institute of Technology concluded 95% of generative AI business pilot programs don’t boost revenue despite $30 billion to $40 billion in investment.
Even OpenAI CEO Sam Altman has concerns.
“When bubbles happen, smart people get overexcited about a kernel of truth,” Altman said in August. “Someone is going to lose a phenomenal amount of money.”
There are warning signs reminiscent of the 2000s dot-com crash or the 2008 real estate meltdown: soaring valuations, vague business models, circular dealmaking and a capital flood. If a collapse happens, it could hit hard in San Francisco, which relies heavily on tech to stimulate its downtown, roughly 75% of which is devoted to office uses. And, the market is still experiencing a severe hangover from the pandemic: Roughly one-third of downtown’s 90 million square feet of office space remains vacant.
Article continues below this ad
But the market, despite Friday’s major drop, is also performing better today than it has in years, thanks to AI’s astronomical rise — real estate firm JLL is tracking more than 150 leases with AI companies that have set up shop in San Francisco since 2020.
Yet, there is no guarantee that the companies currently driving demand for offices in the city will be around five years from now, and lease deals forged today are usually of short duration. Though, San Francisco officials believe AI’s momentum will help reverse its economic slide even if a downturn occurs.
Mayor Daniel Lurie’s office has repeatedly expressed commitment to amplifying the city’s image as an AI innovation hub. The focus has been, among other things, on public safety and reforming the city’s housing and business permitting processes.
While these efforts seemingly meet the moment, there are concerns about whether AI offers a lasting fix. San Francisco’s tech sector embraced remote work faster than any other during the pandemic. Now, even proponents of AI’s momentum question whether the city, and its real estate market, is leaning too heavily into another boom — and risking the same bust cycle all over again.
“I meet many founders now who have basically no revenue, no product and, sometimes, not even a team — it’s just them — and not even a (pitch) deck,” said Henry Shi, who co-founded travel and finance startup Super.com in 2016. “They are raising $5 million, $10 million, $30 million-plus for the initial round. It’s quite incredible, actually.”
Article continues below this ad
In the previous decade’s boom, founders with proven records and prior successful startups could raise blockbuster funds. But now engineers who left established AI companies are easily getting hundreds of millions of dollars for their first startup, without concrete products. When there is a product, it may not be original — such as the multiple startups working on AI receptionists for dentists, Shi said.
“It is a bubble but also it may not matter,” Shi said. Venture capitalists are used to betting on a hundred startups and watching the vast majority fail, in hopes of finding that one unicorn that becomes worth billions. In the world of AI, that one phenomenal success could be worth even more.
“You could have 1,000 scams, but if one of them actually becomes a trillion dollar company then it’s fine. Before (AI) that wasn’t really possible,” he said.
Artificial intelligence conferences have become common events in San Francisco, with several of the gatherings taking place throughout the year.
Lea Suzuki/S.F. Chronicle
The Bay Area has attracted about 75% of the country’s venture capital funding for AI since 2019, according to research firm PitchBook. As of last week, $95 billion has been raised in 2025, which is already the highest annual amount on record and double last year’s total.
Article continues below this ad
A huge portion of that was OpenAI’s $40 billion funding round in March. Meanwhile, non-AI companies aren’t seeing as much investment liquidity amid volatility around tariffs and interest rates and a slow pipeline for initial public offerings.
“There’s a tale of two markets — AI and everyone else,” said Emily Zheng, a senior venture capital analyst at research firm PitchBook. “Concerns of a bubble are justified because we’re seeing such eye-watering valuations.”
And as the industry accelerates, will the immediate benefits to cities like San Francisco — investment, innovation and economic growth — be enough to offset long-term costs like job displacement, soaring housing prices and deepening inequality?
Even some real estate insiders described leasing trends by AI companies as a double-edged sword: While driving activity, some have predicted that their success will likely eliminate most entry-level white collar jobs. The canary in the coal mine appears to be the city’s largest private sector employer, Salesforce. The company’s CEO, Marc Benioff, said last month that 4,000 customer support workers have been replaced by AI. The company didn’t provide comment on a potential AI bubble.
“I’m a skeptic, having seen these bubbles in the past. A lot of wealth is created, but a lot of wealth is destroyed as well,” said economist Ken Rosen, who is the chairman of the Berkeley Haas Fisher Center for Real Estate and Urban Economics. “My experience with previous things like this is: The promise is much greater than the reality. But, there is real stuff here.”
For Rosen, there is “no question” that an unprecedented amount of funding is going into AI “without any assurance that the return is going to be there for at least a couple of years.”
Rosen is tracking what he calls an “AI infrastructure boom” — “AI enablers,” like companies building data centers and providing other support for research and development, are making money, he said. But he is not yet seeing an “AI usage boom.”
“The question is, are enough people using AI in a way that creates revenue or generates productivity? The evidence is pretty sparse at this point,” he said.
Egan, San Francisco’s economist, noted the data center splurge is largely happening outside of the Bay Area, in places where energy costs, land prices and taxes are lower. But such projects also don’t bring as much tax revenue and jobs as do offices full of engineers, he said.
California Gov. Gavin Newsom held a press conference at the Google office in San Francisco in August to announce new AI partnerships.
Anna Connors/S.F. Chronicle
Egan believes that the data center projects outside California would be cancelled first before engineering jobs are cut in San Francisco, if a bubble pops. In that sense, the city could be less vulnerable to a downturn than those in the midwest and more rural areas.
He is more concerned about a broader recession, combined with an AI bubble bursting.
“It’s more about the national economic problems hitting a city that’s still kind of precarious due to the weak recovery we’ve had from COVID,” Egan said.
The good news is that the city’s business taxes are shooting up, he said. A major reason is AI spending mainly fueled by public “Magnificent Seven” tech titans Apple, Microsoft, Google, Amazon, Meta, Nvidia, and Tesla, plus the private OpenAI, which was just valued at $500 billion.
“They invest in each other, drive up each other’s valuations. They’re sort of selling to each other a lot,” Egan said, which “leads to a sort of feeding frenzy that leads to an investment … inflow,” as well as a gap between future revenue compared to spending.
But Egan sees an upside to such concentration: All seven tech titans already generate a great deal of profit through their core businesses — advertising, iPhones, cloud storage, online shopping, chips and cars. They have billions to burn on AI investments, and if they don’t see much return, they still have profit engines to fall back on, which could soften the blow to San Francisco if the bubble pops, Egan said. He also believes the titans have already cut jobs in response to AI and overhiring and sees layoffs slowing.
It’s far from certain that AI will churn out money like iPhones or social media ads, which fueled earlier booms, according to Egan.
“There’s no money in the ChatGPTs and the Claudes of the world,” he said. “The consumer-facing things are just commodities.” (Both OpenAI and Anthropic, the companies behind those two chatbots, have lost billions of dollars annually.)
However, Christina Melas-Kyriazi, a partner at venture capital firm Bain Capital Ventures, said that she is already seeing “large amounts of revenue, but also value” being created, even if companies are not yet profitable. For her, value creation is “the important metric.”
“It’s not just revenue growth — what value are you creating in large verticals like coding, customer support and enterprise search, and then health care, even finance?”
She defined the AI bubble as prices that are “above fundamental value levels.”
“So far, we actually don’t see that,” she said. “The exuberance is in many cases justified because there are real enormous profits at the core.”
“Long term, I think that in a lot of markets, there’s a strong reason to basically capture value and demand early and sort of subsidize getting to scale, because once you capture customers and you capture attention, and you’re deeply integrated into their workflows, you can raise prices,” she said. “I think that’s a strategy that a rational, smart startup founder would take.”
Legal AI startup EvenUp just raised $150 million on a valuation of $2 billion, double its value a year ago. Employees have grown from 100 to 600, and while a portion of those jobs are remote, the company opened its first San Francisco office this summer, stating at the time that it can’t “win” without a presence in the city. EvenUp helps attorneys for injury victims by streamlining the legal process and was founded in 2019, years before the AI hype — leading to three rejections from prestigious incubator Y Combinator, said CEO and founder Rami Karabibar.
Karabibar said his team’s experience was the “reverse” of many seed rounds today: The startup had to prove itself to investors before securing major funding.
“Once we actually launched the product, and once folks saw how much revenue we were adding every month, every quarter, it became really obvious that this thing is here to stay,” Karabibar said. “It’s still early days in generative AI, which is one of the reasons why our investors really doubled down. There is a really big opportunity ahead of us, and we are only getting started.”
Despite alarm bells over an AI bubble, there’s no sign of investment and megadeals slowing.
Rosen predicts that San Francisco will continue to benefit from the “upsurge” over the next couple of years, but is convinced that there will be a reckoning: “When the correction occurs, we will all pay the price in terms of job losses and leases ending.”
When that will happen is anyone’s guess.
“If you think of it as a nine-inning game, we are probably in the third or fourth inning,” he said. “We’re on the upslope right now — so take advantage of it.”