AI Leasing Race Heats Up Between New York And San Francisco

The race for AI tenants is turning into a two-city contest.

San Francisco and New York have both landed a series of large leases from artificial intelligence companies this year, with some of the biggest players expanding their footprints in both markets.

Anthropic is leading the charge on both coasts. The AI giant signed a 466,000-square-foot full-building lease at AEW Capital Management’s 330 Hudson Street in July, the largest AI lease in New York City so far this year, according to CompStak data. In San Francisco, it inked a 413,000-square-foot full-building lease at 300 Howard Street in February, followed a few months later by another 250,000-square-foot lease at 500 Howard Street.

That gives Anthropic more than 1.1 million square feet of new space across the two markets this year alone, per CompStak. The 300 Howard Street lease had a starting rent of $87 per square foot, per CompStak. The average starting rent for AI tenants so far this year in the combined San Francisco and Bay Area markets is $77 per square foot.

Anthropic’s rent in New York was not disclosed, but the average asking rent for AI tenants so far in 2026 is about $82.83 per square foot, per CompStak. But many are paying more. Tech companies inked 69 leases with rents over $100 per square foot since the beginning of 2025, per JLL data.

“Tech was drawn to Midtown South because it was cool and cheaper,” said JLL’s Kevin Kelly. “Now, due to a combination of demand for space and people being aware of what Midtown South has to offer, there’s no discount on what they’re taking down.”

AI tenants in San Francisco have leased more square footage in this year’s biggest deals. Taken together, the five largest AI leases in San Francisco and the Bay Area totaled about 1.38 million square feet. New York’s five largest deals amounted to roughly 1.08 million square feet.

In the largest West Coast deals, OpenAI leased 282,000 square feet at 1800 Owens Street, while Sierra AI took 258,000 square feet at 185 Berry Street. Databricks signed for another 180,000 square feet at 100 Altair Way in the Bay Area, according to CompStak.

New York’s biggest deals after Anthropic include Ramp’s 285,000-square-foot extension and expansion at 28-40 West 23rd Street, Tennr’s 125,000-square-foot lease at 345 Hudson Street and EliseAI’s 109,000-square-foot deal at 401 Fifth Avenue. Legora rounded out the top five with a 98,000-square-foot lease at 11 Madison Avenue, per CompStak.

The numbers offer a snapshot of the competition between the country’s two biggest markets for a generation of AI companies that are rapidly scaling their office footprints. In San Francisco, AI companies have accounted for 30 percent of leasing activity, or roughly 10 million square feet, since 2023, according to CBRE data. In Manhattan, AI companies leased 4.1 million square feet during that period, per CBRE.

But the Bay Area is far more reliant on AI to drive office demand. AI companies accounted for nearly half of all office leasing by square footage in San Francisco so far in 2026, said CBRE’s Colin Yasukochi.

“Many of the larger AI companies in New York are actually headquartered in the San Francisco Bay Area,” he said. “When they reach the sort of scale that they are in San Francisco … and they need to hire across the spectrum, but they’re still focused on hiring the innovative technical talent, New York has the largest supply.”

AI companies also want to be close to the customers and industries they’re trying to sell to, Yasukochi said. New York’s concentration of finance, media and other industries gives companies access to workers who understand those businesses.

“Having some sort of knowledge of other industries, like finance, like media, and just understanding that, then they can build products and services from AI that relate more directly to that, as opposed to tech companies building it for other tech companies,” he said.

The race isn’t just about how much space AI companies are taking. It’s also about a finite pool of highly specialized workers, said JLL’s Kelly. Neither New York nor San Francisco has enough workers with the specialized skill sets AI companies are seeking, so companies like Anthropic are expanding in both markets.

“When they’re looking at it in terms of discrete skills, a lot of these companies … are growing in multiple places at the same time because there’s simply not enough people to hire at the rate that they want to go for those elite skills,” Kelly said.

The talent pipeline is shifting, too. Kelly said that around 2015 and 2016, computer science graduates from schools like Michigan, Purdue and Wisconsin were much more likely to head to the Bay Area than New York. That pipeline has since shifted toward New York, and AI companies have taken notice.

“They’re not thinking about low-cost hubs. They’re thinking about ‘I need the best and brightest, and I’m going to scale them up in these core markets where these skill sets are most significant,’” Kelly said.

The companies aren’t necessarily leasing only for the workers they already employ. Kelly said JLL’s analysis of headcount against occupied square footage suggests that 50 to 60 percent of the space AI companies are taking is earmarked for employees they have yet to hire.

“They are planning for such a significant ramp-up, so these are really big-time costs they’re taking on,” he said.