Big Tech's green credentials face a new challenge: activist investors want answers on AI energy use. Amazon, Google, and Meta-companies that promised net-zero emissions-are now building massive AI data centers that could derail those goals.
Shareholder resolutions focus on the environmental costs of AI. Training and running large models requires vast amounts of electricity and water. AI currently consumes enough power for 7 million US homes, a figure projected to reach 22% of US households by 2028. North American data centers used nearly 1 trillion liters of water in 2025, equal to New York City's annual consumption.
Microsoft offers a benchmark: its data-center electricity use may double by 2030, even though it aims to be carbon-negative. About 97% of its emissions come from its value chain. Independent assessments rate Amazon and Google's climate pledges as “low integrity,” with Apple scoring only “moderate.”
Investors argue this is greenwashing. Companies marketed themselves as climate leaders while planning capital expenditures that increase energy footprints. The push is not fringe: institutional investors with significant positions are joining, citing financial risks if power costs spike or regulations tighten.
The AI arms race means tens of billions are spent annually on infrastructure. Each new data center is a long-term energy commitment. Historically, environmental resolutions rarely pass with binding majorities, but support is growing.
For investors, the calculus shifts. The AI bull case assumes manageable costs. Activist investors flag a scenario with energy constraints, water scarcity, and regulatory pressure making AI buildout slower and more expensive. That gap between expectation and reality is where risk lives.