San Diego Pauses New Data Centers While It Works Out Power, Water and Neighborhood Rules

San Diego Pauses New Data Centers While It Writes the Rules

San Diego City Council unanimously ratified a temporary data center moratorium on October 27, following an executive order from Mayor Todd Gloria on October 22. The pause lets the city assess electricity demand, water supplies, infrastructure costs and neighborhood impacts while it develops regulations for data center development. The Planning Department must set up a stakeholder working group, including technology, community, utility, water, housing and land use representatives, and departments must submit preliminary recommendations within 180 days.

This is a local decision with a national pattern behind it. Cities across the US are discovering that data centers, which were once welcomed as quiet, job-light tax revenue, now raise hard questions about power grids, water use and land prices. San Diego is choosing to pause and write rules before approving more projects, rather than approving projects and dealing with the consequences later. That order of operations is the story.

What the San Diego Data Center Moratorium Covers

The moratorium is temporary, not a permanent ban. It pauses new data center development within the city's jurisdiction while planners study the impacts and recommend regulations covering location, construction, operation and expansion. The mayor's order framed the concern plainly: residents should not absorb unnecessary costs or neighborhood impacts from development they did not ask for.

The 180-day clock for preliminary recommendations gives the process a deadline, but not a quick one. Six months is enough time for a working group to gather utility data, hear from residents and draft rules, but it is also long enough for developers to shift projects to neighboring jurisdictions with looser rules. That tension, between careful local regulation and regional competition for investment, runs through every data center debate in the US right now.

Why Power and Water Dominate the Debate

Data centers use electricity at a scale that can strain local grids, and they use water for cooling, especially in hot climates. San Diego, in Southern California, faces both constraints at once: a grid that must also support electrification of transport and buildings, and water supplies that are always a planning concern. A single large data center can use as much power as a small town, and its load is constant, day and night, which makes grid planning harder than handling a factory with variable shifts.

The cost question is just as important as the resource question. If a utility builds new transmission or generation capacity for data centers, who pays? If residents pay through higher rates, the data center becomes a subsidy in reverse. If developers pay, projects become more expensive and some may not proceed. Cities are now trying to answer that question in regulation, rather than discovering the answer in rate cases years later. San Diego's working group, with utility and community representatives at the same table, is designed to force that conversation early.

San Diego Is Not Alone in Pausing

San Diego follows a wider pattern of local pauses and zoning controls. Digital Watch reporting noted 45-day pauses in San Francisco and Oakland, and a six-month moratorium on site plan approvals in Calvert County, Maryland. Beyond the US, a petition in Scotland calls for a moratorium on hyperscale data center approvals until environmental and community impacts are assessed and planning safeguards are strengthened. The pattern is consistent: local governments are using pauses to buy time to write rules that did not exist when the data center boom began.

The industry response is also consistent. Operators argue that data centers bring investment, tax revenue and infrastructure, and that modern facilities are far more efficient than early ones. Both points can be true at the same time as residents' concerns. A data center can be efficient per unit of compute and still strain a local grid if enough of them cluster in one area. That is why location rules, not just efficiency standards, are central to the new regulations cities are writing.

What This Means for the AI Buildout

San Diego's pause does not stop the AI buildout. It slows one city's approvals while the buildout continues elsewhere, including in places with cheaper power and looser rules. The CoreWeave expansion into Denmark and Spain's softened data center rules show the other side of the same coin: some jurisdictions are competing to attract data centers, while others are pausing to control them. The result is a patchwork, where the same project can face very different rules a few miles apart.

For cloud customers and AI companies, that patchwork means longer timelines and more local politics in capacity planning. A data center that was once approved in months may now take years, with community hearings, grid studies and water assessments along the way. That delay is a cost, but it may also be a necessary correction. The first wave of data center approvals happened before AI demand exploded, and the rules written for that era do not fit the scale of today's projects.

What to Watch in the Next 180 Days

Watch for the working group's composition and its preliminary recommendations, due within 180 days. The key questions are: where in the city can data centers be built, what power and water conditions apply, who pays for grid upgrades, and what community benefits, if any, are required. Watch also for whether developers challenge the moratorium in court, or simply move projects outside city limits.

San Diego's experiment will be watched by other cities facing the same pressures. If it produces clear, fair rules that still allow well-sited projects, it could become a template. If it produces a long, uncertain process that pushes all development elsewhere, it will be cited by industry as a warning. Either way, the era of approving data centers as if they were ordinary warehouses is ending, and San Diego just made that official.