← Back to feed

Residents of Gilroy, California discovered an Amazon data center was already under construction in their city without a community vote or public input process. Amazon used a 45-year-old California planning provision that allows streamlined approval for certain projects, effectively closing the window for residents to object before ground broke.

Gilroy sits roughly 30 miles southwest of San Jose, in Santa Clara County’s agricultural corridor. The data center, which Tom’s Hardware first reported on, drew widespread local surprise: residents say they had no knowledge of the project until construction was underway.

The Mechanism

California’s streamlined environmental review process was enacted in the late 1970s and early 1980s to reduce permitting delays on certain categories of development. Projects that meet specific criteria can bypass the public comment periods required under CEQA (the California Environmental Quality Act). Developers use it for speed; critics have long argued it was designed for routine commercial builds, not facilities with the power and water footprints of AI data centers.

Amazon’s Gilroy project qualified under the streamlined pathway. A community vote that had been discussed locally was rendered moot once the procedural track was invoked. Residents were informed not before the decision but after the foundation work started.

The Pattern

Amazon is not alone. Across the country, AI data center developers have used quiet acquisition strategies, state-level preemption, and procedural mechanisms to get shovels in the ground before opposition can organize.

Earlier this year, Stack Futures reported that community opposition had blocked $130B in AI data center projects nationally, making it the fourth major constraint on expansion after power, land, and permitting speed. Amazon’s Gilroy move reflects an industry-wide shift in tactics: instead of fighting public hearings, developers are finding the procedural exits before hearings are scheduled.

Santa Clara County has seen significant data center concentration near San Jose. Gilroy, historically a garlic-farming town that has evolved into an agricultural service hub, represents a second tier of California geography now entering the data center expansion zone.

What This Means

The 45-year-old provision Amazon used was not written with gigawatt-scale AI infrastructure in mind. It was designed to accelerate routine commercial permits. Its applicability to large data centers, with their demands for dedicated power substations, high-volume water cooling, and 24-hour industrial activity in residential-adjacent areas, is being challenged in several California municipalities.

Gilroy’s situation is likely to add fuel to calls for California to explicitly carve out AI data center projects from streamlined review eligibility. Oregon and Texas have already amended their CEQA equivalents in the opposite direction, accelerating data center approvals further. California faces competing pressure: it wants the economic activity, but its environmental review framework was written to give communities a voice.

Amazon has not commented publicly on the Gilroy project scope, timeline, or capacity.