By Right
A law written to stop one kind of local veto turns out to have no way of telling it apart from a different kind of local veto — the kind protecting a watershed nobody at the hearing gets to have, because there isn't one.
Imperial County, California, sits over a groundwater basin that took millennia to fill, in a region already contesting every acre-foot of Colorado River allocation it has left. In early 2026, a developer proposed a data center there just under a million square feet — big enough, by the county's own estimate, to draw more electricity than the entire county used in all of 2024, and to consume roughly 750,000 gallons of water a day cooling the servers inside it. The county's planning department determined the project qualified for approval by right: the underlying parcel was already zoned industrial, so no discretionary hearing, no environmental impact report, no vote weighing the water table against the tax revenue was legally required at all. The project could simply proceed, the way a warehouse or a machine shop proceeds on land already zoned for one.
What happened next inverted the normal shape of a land-use fight so cleanly it's worth sitting with before getting to the mechanism that produced it. The city of Imperial sued the county, arguing the ministerial approval was improper. Residents, environmental groups, and the county's own supervisors spent months organizing against a project none of them had gotten an ordinary hearing to weigh in on. The Board of Supervisors eventually passed a 45-day moratorium on new data center approvals, unanimously, and stood up a committee to actually study what the county's land and water could sustain. And then the developer sued the county — not to defend a permit already earned through public process, but to keep a freeze from interrupting a project the law had never required anyone's discretionary approval for in the first place. The community was cast as the plaintiff seeking to slow something down, using litigation and moratoriums to claw back a review process that would, in almost any other context, have simply been the ordinary starting point.
The mechanism, traced back to its actual origin
That inversion isn't a fluke of Imperial County's specific zoning code. It's the direct, foreseeable consequence of a genuinely well-intentioned state law solving an entirely different problem. California's Senate Bill 35, passed in 2017, created a ministerial, CEQA-exempt approval pathway for qualifying housing projects in cities that had failed to meet their state-assigned housing targets — a direct response to decades of local governments using discretionary hearings, environmental review, and endless procedural delay to block the construction of homes their own constituents desperately needed. The bill's entire theory of the case was correct: local discretion had become a tool for exclusion, not deliberation, and removing it in narrowly defined circumstances was the only way enough housing was ever going to get built.
The category SB 35 created — a class of project that clears zoning and gets ministerial, non-discretionary approval with no CEQA review required — turned out not to know it was only supposed to apply to housing. Fourteen data centers have been approved in California since 2018 under exactly this logic, developers arguing successfully that a data center on already-industrial-zoned land is no different, procedurally, than the affordable apartment building the ministerial pathway was actually built for. Nothing in the legal category distinguishes "a use this parcel was always going to see eventually" from "a use whose water draw alone exceeds what the surrounding basin can indefinitely sustain." The mechanism doesn't ask that question, because it was built to stop a different question — "is this town going to keep finding pretexts to block housing forever" — from ever being asked again. A tool built to remove one kind of illegitimate local veto turns out to have no way of telling that veto apart from a legitimate one, because the law was never written with the second kind in mind at all.
The pattern isn't confined to one bill, either, or to a rural desert county with an obvious water crisis already in the news. On August 18, 2026, the San José City Council rezoned 35 acres of IBM's Almaden research campus from site-specific Planned Development to plain Industrial Park zoning. The vote's practical effect had nothing to do with what IBM currently does on that land and everything to do with what the rezoning makes newly possible: Industrial Park zoning is exactly the designation that qualifies a project for the CEQA exemption Senate Bill 131 — a 2025 state law aimed at attracting "advanced manufacturing" — grants to qualifying projects built on industrially zoned land. Residents wrote in asking the council to pull the item off the consent calendar and debate it in the open rather than wave it through with the rest of the routine business; District 10's own councilmember told constituents, in substance, that no amount of community input was going to change the vote. Whatever eventually gets built on that hillside — hazardous materials handling, wastewater, truck traffic, the water and power draw any advanced-manufacturing tenant would bring — now has a real chance of arriving with no environmental impact report, no alternatives analysis, and no requirement that a single significant impact ever get mitigated, because the zoning designation that unlocks all of that exemption was itself approved as a consent-calendar formality, a full legislative step removed from the facility whose impacts will actually land on Almaden Valley.
That's the same mechanism running one step further upstream. SB 35 lets a category of project skip review once it already sits on the right zoning. SB 131 shows the same defect can be triggered proactively — a city council can simply vote the zoning designation into existence first, on a technical-sounding land-use item nobody outside the neighborhood was watching, and the CEQA exemption follows automatically once the label is attached. Two different bills, aimed at two different goals — housing supply in one case, manufacturing investment in the other — produced the identical structural hole: a legal category built around what a parcel is zoned for, with no mechanism at all for asking what a specific project actually does to the water table, the grid, or the neighborhood downhill from it.
What a community actually has standing to claim
The honest version of this essay has to ask the harder question directly: does a community actually have a legitimate interest in a project sited entirely on privately owned land the zoning code already permitted, or is "we didn't get a say" simply the same exclusionary instinct SB 35 was built to stop, wearing a watershed's name instead of a neighborhood's?
The answer turns on exactly the distinction Land as Road already made about land generally: some consequences of a land-use decision stay entirely inside the parcel line, and some don't, and the ones that don't are where a community's claim stops being NIMBYism and starts being a legitimate accounting of a cost nobody priced. A single data center's aesthetics, its traffic, even its noise are largely internal to the site and its immediate neighbors — exactly the register of complaint SB 35 was right to stop treating as a permanent veto. A groundwater basin and a regional electrical grid are not internal to the site at all. They're shared commons whose total capacity was never any single landowner's to spend down unilaterally, in the same sense A Meadow of Sacred Cows argued a location's value is never entirely its titleholder's own creation. A parcel zoned industrial gives its owner the right to build a warehouse on it. It was never plausibly meant to give a single private developer the unreviewed right to draw down an entire county's annual power budget or claim three-quarters of a million gallons a day from an aquifer every other water right in the basin also depends on — and the fact that the zoning code's language doesn't distinguish those two things is the actual defect, not the fact that a community objected.
It's worth being clear-eyed about who's actually on the other side of that unreviewed right, too, because it's rarely a single local developer risking their own capital. BlackRock, Global Infrastructure Partners, Microsoft, and the Abu Dhabi-backed investment vehicle MGX launched a joint partnership in 2024 aimed at mobilizing $30 billion in private capital — as much as $100 billion once debt financing is layered in — specifically to build data centers and the power infrastructure behind them, and that same consortium moved to acquire the entirety of Aligned Data Centers in a deal valued at roughly $40 billion. This is Mancur Olson's concentrated-benefit side of the ledger at a scale ordinary land-use disputes rarely see: a handful of the largest asset managers and sovereign investment funds on earth pooling capital behind a single facility, set against a rural county or a single neighborhood whose only leverage is a moratorium vote or a lawsuit filed after the fact. The mismatch in scale is exactly why "by right" approval mattered enough to litigate over in the first place — a community organizing a 45-day freeze is not a contest between equals to begin with, even before the legal category removes its usual seat at the table entirely.
The reporting bills are the "who pays" question, arriving late
California's 2026 legislative response is worth reading as an attempt to draw exactly that line rather than simply hand local discretion back wholesale. SB 887 would end categorical CEQA exemptions specifically for data centers, requiring at minimum an initial environmental study, while still offering an expedited-review track for projects that voluntarily commit to zero-carbon power, water-efficient cooling, and community benefits agreements — a bet that most developers will take the fast lane once it comes with real conditions attached, rather than the current arrangement, where no lane requires any conditions at all. AB 1577 and AB 2619 would require data centers to report their actual electricity and water consumption to state regulators monthly, including a new "indirect water use" metric counting the water consumed generating the electricity a data center draws, not merely the water run through its own cooling towers. SB 886 would put data centers drawing 25 megawatts or more on a dedicated electrical tariff, with a fifteen-year exit penalty attached — a direct answer to the "who pays?" question Who Pays? asked about ledgers and A Meadow of Sacred Cows asked about sea lanes, aimed specifically at stopping a hyperscale facility's power draw from being quietly absorbed into everyone else's electric bill the way it currently can be.
None of these bills restore the old discretionary hearing SB 35 was built to eliminate. That's the right instinct, not a compromise forced by politics. A return to unlimited local discretion would just resurrect the exact tool that spent decades blocking the housing California actually needs, now redirected at data centers instead — trading one illegitimate veto for a different illegitimate veto rather than fixing the actual defect. What these bills attempt instead is a narrower, criteria-based review keyed specifically to the commons a project actually draws on — water, grid capacity, emissions — leaving the ordinary internal-to-the-parcel questions exactly as ministerial as SB 35 already made them. Whether it lands as narrowly as intended is still being litigated in real time, but the theory of the fix is sound: the problem was never that a category of project got ministerial approval. It's that the category never distinguished a housing unit, whose externalities mostly stop at its own walls, from a hyperscale data center, whose externalities reach an entire regional watershed and grid.
What's actually being renegotiated, and how fast
Every other sacred cow this site has examined took decades to even become contestable — total fire suppression ran unquestioned for the better part of a century before the science caught up with it; the naval protection of global shipping has never been seriously put to a vote at all. Data center siting is the unusual case running the opposite way. More than 200 bills addressing data centers were introduced across all fifty states in 2025 alone, more than forty of them already enacted; local campaigns cancelled or delayed north of $170 billion in projects nationally in the first quarter of 2026 alone, water use cited as a leading concern in most of them. Whatever gets settled in California this year will settle fast, by the standard of every other arrangement this site has covered, precisely because the underlying resource — water, in a region already fighting over what's left of it — makes the externality impossible to keep invisible the way a subsidized sea lane or a suppressed wildfire risk can stay invisible for a century.
That speed doesn't make the underlying question any less real. A data center sited entirely by right, with no hearing anyone attended and no vote anyone cast, drew down water and power that belonged, in the only sense that actually matters, to everyone else drawing on the same basin and the same grid — and the only reason Imperial County's residents ever got a hearing on any of it was that they organized hard enough, fast enough, to force one into existence after the fact, using a moratorium and a lawsuit to manufacture the review process the law had already decided they didn't need. A community's interest in a watershed was never actually in dispute here. What was in dispute is whether the law had a category capable of recognizing it — and for the fourteen data centers already approved before anyone thought to ask, it didn't.
Sources: CalMatters, Data centers are guzzling California's water. We have no idea how much.; CalMatters, This million-square-foot data center would be the biggest in the state; KPBS, After months of public pressure, Imperial County passes temporary moratorium on data centers; iNewsource, AI data center plan energizes opposition in California desert; iNewsource, California's largest data center plan rouses Imperial County; Calexico Chronicle, Imperial's CEQA Claims Are Insufficient to Challenge Data Center; Mayer Brown, Proposed California Legislation Aims to Reshape Land Use Approvals for Data Centers and Renewable Energy and Storage Projects; Senator Steve Padilla, Bills Protecting Ratepayers from Data Center Costs and Incentivizing Sustainable Tech Infrastructure; Wikipedia, Opposition to AI data centers; San José Inside, Almaden IBM Rezoning: When CEQA Disappears, So Does Your Voice; BlackRock, AI Infrastructure Partnership; Global Infrastructure Partners, AI Infrastructure Partnership (AIP), MGX, and BlackRock's Global Infrastructure Partners (GIP) to Acquire All Equity in Aligned Data Centers.