Front Burner The true cost of AI data centres
Mar 24, 2026
Ellen Thomas, an investigative reporter at Business Insider who has tracked the AI data centre boom, discusses the scale and lifespan of massive server farms. She covers water and power strain, why AI cooling uses so much water, local pushback and cancelled projects, and hidden environmental costs like wetlands loss and backup fuel emissions.
AI Snips
Chapters
Transcript
Episode notes
AI Workloads Multiply Energy And Cooling Needs
- AI workloads are far more energy intensive than typical cloud tasks, roughly 10× the processing energy, generating extreme heat.
- That heat drives heavy water-based cooling systems, explaining large water consumption in AI-focused centres.
Grassroots Organizing Can Quickly Halt Projects
- Local opposition often begins with neighborhood groups or residents living next to proposed sites and can escalate to organized campaigns.
- Ellen Thomas cites Memphis where community organizing around Elon Musk's XAI site led to heated health department hearings.
Local Pushback Is Causing Billions In Delays
- Community pressure has real impact: Data Center Watch logged $98 billion of projects delayed or canceled in a few months in 2025.
- Thomas says companies avoid delays and will relocate if opposition is strong.