More than a dozen investors are pressuring Amazon, Microsoft, and Alphabet's Google to provide detailed data on water and energy consumption at their U.S. data centers, Reuters reported today. The pressure comes as all three companies have recently scrapped multibillion-dollar data center projects following community opposition, and as North American data centers consumed nearly 1 trillion liters of water in 2025, according to market research firm Mordor Intelligence.

Water usage is still a key concern for AI data centers. Some estimates say that generating just 100 words with OpenAI's GPT-4 consumes three bottles of water. Last year, another study suggested that AI data centers use more water annually than people drink bottled water, globally.

Article continues below

All four major hyperscalers have adopted closed-loop cooling systems that use substantially less water than traditional evaporative methods, but the level of disclosure varies.

Data center water consumption involves more than closed-loop cooling, as well. The majority of water a data center consumes is indirect; i.e. for generating the electricity the data center will use. A 2024 study found that data centers (in general, not specifically AI data centers) consumed about 800 billion liters of water indirectly. Another 2025 study found the vast majority of water consumption happens offsite an AI data center.

Investors want site-level data to assess operational risks more accurately. "We haven't seen them disclosing enough about their water consumption (and the) impact on the local community," Jason Qi, lead technology analyst at Calvert Research and Management, told Reuters.

Josh Weissman, director of infrastructure capacity delivery at Amazon, told Reuters the company is "increasingly disclosing site-specific water consumption data where we operate." An Amazon spokesperson added that the company was committed to being a "good neighbor" and was investing in efficiency and water reduction efforts. A Microsoft spokesperson said environmental sustainability was "a core value." Google declined to comment, and Meta did not respond to Reuters' request.

The backdrop of this is the alarming rate of AI data center expansion. Last week, Bloomberg reported that nearly half of planned U.S. data center builds for 2026 have been delayed or canceled, largely due to infrastructure demands. Although there's a clear call from some investors for sustainability, there's also a question of practicality when new projects are being funded so quickly. The capital expenditure of companies like Microsoft, Amazon, and Meta has shot up amid the AI boom, even if some projects will never actually be completed.

Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.