Amazon is facing accusations that it sidestepped a public vote required for large development projects in Gilroy, California, by invoking a zoning provision that dates back roughly 45 years, allowing a massive AI-focused data center to move forward without the standard community comment period, according to a report from Tom’s Hardware.
The report says Gilroy officials and Amazon relied on an older land-use classification to argue the project could proceed through an administrative or ministerial approval process rather than the discretionary review that typically triggers public hearings and a community vote. Residents in the Santa Clara County city, roughly 80 miles southeast of San Francisco, say they were effectively locked out of the window during which the public would normally be able to formally weigh in.
How the Loophole Works
Many California cities, including Gilroy, maintain zoning codes and specific plans that predate the current wave of hyperscale data center construction by decades. When a proposed project can be characterized as consistent with a pre-existing land-use designation, local governments can sometimes classify approval as ministerial rather than discretionary. Ministerial approvals are largely a rubber stamp against a fixed checklist and do not require the same environmental review, public hearings, or opportunities for community input that discretionary approvals demand under state law.
Critics say this is precisely what happened in Gilroy: a rule written long before anyone anticipated a facility of this scale and power draw was used to classify a modern AI data center as routine, administratively approvable construction. That distinction matters enormously to residents, because discretionary review is typically the only point at which a city council vote or public hearing is legally required.
Community advocates argue that using decades-old zoning language to bypass modern environmental and public-input requirements defeats the purpose of those requirements altogether, since the rules were never written with gigawatt-scale AI infrastructure in mind.
What Residents Are Saying
Gilroy is a city of roughly 60,000 people known historically as an agricultural hub. Residents who spoke with Tom’s Hardware said they learned about the scale of the planned facility only after key procedural decisions had already been made, leaving little practical opportunity to organize opposition or request changes before construction commitments were finalized. That mirrors a pattern playing out in numerous US communities where hyperscale data centers have moved in with limited public notice.
Concerns commonly raised by residents near large data center projects include:
- Strain on local electricity grids and the potential for rate increases passed on to residential customers
- Heavy water usage for cooling systems, particularly in regions already facing drought pressure
- Noise from constant cooling equipment and backup generators
- Loss of agricultural or open land to industrial-scale infrastructure
- A lack of transparency about which company is actually building a facility, since data center projects are frequently filed under shell company names
Part of a Broader Pattern
The Gilroy dispute lands amid a nationwide boom in data center construction driven by demand for AI training and inference capacity. Amazon, Microsoft, Google, and Meta have all announced multibillion-dollar data center campuses over the past two years, and local governments in Georgia, Virginia, Texas, Indiana, and elsewhere have faced similar criticism for approving projects with minimal public process, often citing economic development incentives and confidentiality agreements tied to the negotiations.
Zoning and permitting fights have become one of the primary battlegrounds where communities can still exert influence over AI infrastructure buildout, since many of the underlying environmental and energy policy questions are decided at the state or federal level. When local governments rely on ministerial pathways or older ordinances to avoid discretionary review, opponents argue it removes one of the few levers residents have to negotiate concessions, such as noise limits, water-use caps, or community benefit agreements.
What Happens Next
It remains unclear whether Gilroy residents or advocacy groups will pursue legal action challenging the ministerial classification, a strategy that has succeeded in some California jurisdictions under the California Environmental Quality Act when courts have found a project does not genuinely qualify for streamlined review. Amazon has not issued a detailed public response to the specific procedural allegations reported by Tom’s Hardware.
The episode adds to a growing list of AI-adjacent controversies drawing public scrutiny this year, from security research exposing hidden vulnerabilities in widely used software to AI systems producing bizarre and misleading claims. As tech companies race to build out physical AI infrastructure at unprecedented speed, the Gilroy case is likely to become a reference point for other communities trying to determine what leverage, if any, they retain once a data center proposal reaches their city planning department.
For now, Gilroy stands as a case study in how infrastructure built for a 2026 technology boom can be authorized under rules written for a very different era, long before anyone anticipated facilities capable of drawing hundreds of megawatts of power to train artificial intelligence models.
