“San Francisco is the global hub for AI innovation,” said San Francisco Mayor Daniel Lurie. It is home to headquarters of major AI companies like OpenAI and Anthropic, but thus far, companies have largely built their data centers in the American Midwest and South.
In an article published on Sept. 18, the San Francisco Chronicle revealed that a real estate private equity firm is working to develop a 160,000 square foot data center on San Francisco’s waterfront, in the Dogpatch neighborhood. The company is eyeing the industrial lot at 1215 Michigan Street for redevelopment. This is a larger-than-average data center, roughly three football fields in size.
The Chronicle found that, although the company, Nestory Park Capital, has not disclosed the location of the prospective project, public records reveal that the company and their lobbyist met with Lurie’s chief of housing and economic development, Ned Segal, to discuss planning and building permits for development at the waterfront site.
However on Sept. 14, Shamann Wilson, supervisor for District 10, proposed a 45-day moratorium (temporary ban) on the development of AI data centers in San Francisco. This was sponsored by several other district supervisors. District 10, which includes Dogpatch and Bayview-Hunters point, already has two active data centers in the Bayview neighborhood.
This follows a larger trend of emergency moratoriums against data center construction enacted by district supervisors across California. The idea is that the pause in development would give local governments time to gauge potential impact on their communities and enact policy to regulate data centers.
Manal Faisal, junior computer science major with a concentration in AI, stressed the importance of where the data center is being placed. “Especially in Bayview, the residents there are probably not getting any type of voice in this, and I think it’s very important … with these data centers taking up a lot of water and resources, and affecting the climate as well. I just think it’s not fair,” Faisal said. “Long-term effects are gonna be really bad too.”
Bayview, a historically Black community and a current lower-income community, already faces disproportionate environmental and health hazards like air quality and toxic waste pollution.
A study by the International Energy Agency found that an average AI data center can use around the same amount of energy as 100,000 households. Yale University reported that the energy emissions from a large data center can actually raise temperatures in its localized area, worsening the urban heat island effect.
Cherry Awad, senior biology major said, “In the past, when they’ve built data centers they’ve built them, maybe, in under-resourced communities that maybe were also not told the full scale of things,” Awad said. “I think the reason they’re trying to keep it so hush-hush is because we are a tech city, and I feel like people know what it entails.”
Lurie and his team have adamantly supported AI development in the city, which they say will generate innovation and problem solving. Ted Egan, San Francisco’s chief economist, told the Chronicle that the AI industry is bringing “an explosion of investment” to the city, but has not yet contributed to increasing employment rates.
“We are the home of AI, and we should lean into it. We should absolutely make sure guardrails are in place, but I don’t think that’s the job of the mayor of San Francisco,” Lurie told KQED.
Despite being a global leader in AI, San Francisco currently does not have a land-use designation or legislative definition for data centers. If not written into the city’s land use code, these oversights leave it vulnerable to data centers quietly developing under the residents’ radar and the city unprepared to regulate them, experts told the San Francisco Standard. San Francisco also does not have requirements targeted at data centers’ water or land usage. Currently, to build a data center in San Francisco, companies usually apply for a conditional use permit with the city’s Planning Commission.
Lucy Skelton, sophomore international business major, thinks the development of the data center will “affect culture and life within San Francisco. I think there’s already a very large influx of people moving here [to] work in tech,” she said. “I think it’ll just take away from a lot of the history and culture that San Francisco has, because people who are from the city and grew up in the bay aren’t gonna be able to afford to live.”
According to a poll conducted by the Public Policy Institute of California, 73% of Californians are opposed to AI development in their communities. Activist organizations in San Francisco are currently organizing protests and petitioning against the development of a large data center in the city.
According to a Deep Mind search of two AI Data Center archives there are fifteen AI Data Centers in San Francisco, 32 in the immediate vicinity and an estimated 150 in the triangle region formed by San Francisco, Alameda and Santa Clara County.
An AI assisted search of the 200-400 Paul Avenue AI Data Center Campus documents eight actively operating data centers along a street zoned for light industry that lies across the street from homes and a block west of the major transit station, Caltrain tracks and open windows of senior apartments on Armstrong and Carroll.
The data center map highlighted as a url confirms the existence of 15 AI Data Centers in San Francisco! I urge the authors to read the information you publish!!