Activist Erin Brockovich has created a public map documenting AI data center projects across the United States. The site lists 33 operating facilities, 44 under construction, and 27 proposed sites, while aggregating 2,716 community‑submitted complaints about energy use, water demand, noise, and e‑waste. The growing opposition—reflected in a recent poll showing 70 % of Americans against nearby data centers—has already prompted 69 local jurisdictions to impose moratoriums.
Erin Brockovich Turns Her Attention to AI Data Centers
)
The activist who helped expose the PG&E water contamination scandal is now focusing on the rapid rollout of AI‑focused data centers. On May 24, 2026, her new website, Brockovich AI Data Center Reporting, went live with a searchable map that aggregates community‑submitted issues from every corner of the United States.
What the Tracker Shows
- 33 operational AI data centers (built and running)
- 44 under‑construction projects
- 27 proposed sites awaiting permits or financing
- 2,716 community‑reported locations with complaints ranging from power spikes to water usage
These numbers illustrate a race that is “unfolding town by town,” as Brockovich wrote on the site. The map visualizes where projects cluster—particularly in the Pacific Northwest, Texas, and the Carolinas—while also highlighting regions where opposition has stalled construction.
Core Concerns Raised by Residents
| Issue | Typical Impact | Example Metric |
|---|---|---|
| Energy consumption | Drives up local utility rates; strains regional grids | Some sites demand >150 MW, enough to power 120,000 homes |
| Water usage | Large‑scale cooling systems can withdraw millions of gallons per day | A 100 MW facility may use up to 5 MGD (million gallons per day) |
| Noise & infrasound | Low‑frequency hum from chillers can affect sleep quality | Reports of 45–55 dB constant background noise in nearby neighborhoods |
| E‑waste | Decommissioned servers generate hazardous waste streams | Average 2 tons of electronic scrap per 1 MW of capacity per year |
| Location risk | Proximity to floodplains or seismic zones raises long‑term reliability concerns |
The data center industry often cites liquid immersion cooling and renewable‑energy contracts as mitigations, but the community reports suggest that those measures are not uniformly applied.
Market Context and Policy Reaction
- A survey released in March 2026 shows 70 % of Americans oppose a data center being built within a 5‑mile radius of their home, up 23 % from six months earlier.
- 69 local jurisdictions have enacted moratoriums on new data center construction, buying time for impact assessments and infrastructure upgrades.
- The White House’s AI‑friendly agenda continues to promote rapid deployment of compute capacity, yet the federal administration has not yet issued clear guidelines on environmental safeguards for AI infrastructure.
These dynamics create a tension between the demand for petaflop‑scale compute—driven by large language model training and inference workloads—and the capacity of regional power and water systems. For example, a single 400‑GPU cluster can consume roughly 12 MW continuously, translating to an additional 100 MW on a typical utility grid when multiple sites are built simultaneously.
Potential Legal and Financial Outcomes
Brockovich’s platform functions as an early‑stage evidence‑gathering tool. While no class‑action lawsuit has been filed yet, the aggregated complaints could serve as the factual basis for future litigation. Developers may face:
- Increased permitting costs as municipalities demand more rigorous environmental impact studies.
- Higher utility rates for local residents if data centers receive preferential grid access.
- Investor scrutiny, especially from ESG‑focused funds that track carbon intensity and water risk metrics.
What This Means for the Semiconductor Supply Chain
AI data centers are the end point of the semiconductor value chain. Their growth fuels demand for high‑performance CPUs, GPUs, and specialized AI accelerators, which in turn pressures foundries to push newer process nodes (e.g., 3 nm and 2 nm) at higher volumes. If community pushback curtails site construction, the capacity bottleneck could ripple back to chip manufacturers, potentially slowing the rollout of next‑generation AI chips.
Looking Ahead
The Brockovich tracker will continue to update in real time as new reports arrive. Stakeholders—ranging from utility regulators to chip fab executives—should monitor the map to gauge where social license may become a limiting factor for AI compute expansion.
For more details on the reporting platform, visit the Brockovich AI Data Center Reporting site.
Comments
Please log in or register to join the discussion