As the artificial intelligence industry expands, so does its ecological footprint. ChatGPT alone uses an estimated 39.16 million gallons of water daily. That is the equivalent of everyone in Taiwan, a country of approximately 24 million people, flushing their toilet once. If the industry keeps growing unchecked, its environmental costs will keep escalating with it.
The vast majority of that water is used to cool the servers that keep large AI language models running. For this reason, data centers are often built near abundant freshwater sources that can be easily accessed for cooling. Most facilities are concentrated in the United States, where tech giants are now facing growing scrutiny for their unregulated water consumption. Canada, home to roughly 20% of the world’s freshwater, is becoming an increasingly attractive alternative. Its vast water reserves and cooler climate draw North American companies seeking to expand their AI infrastructure north of the border.
The increase in water usage exacerbates the problem of dwindling freshwater supply. Climate change exacerbates this scarcity further by creating drought and changing precipitation levels, which disrupts natural patterns of evaporation and rainfall. Additionally, heavy rainfall after drought carries the risk of introducing acute gastrointestinal illness (AGI) into the water supply, by picking up such a high volume of sediments that it overloads filtration systems’ ability to remove pathogens and bacteria. As tech companies continue their unabated consumption of scarce resources, it exacerbates these effects of climate change.
Canada’s growing AI industry depends on its freshwater advantage, but without mandatory water-use reporting and legislation, this advantage could become a liability.
The cooling of data centers is a complicated process. Water can come from three sources: blue (surface water and groundwater), municipal supply, and gray sources (purified reclaimed water). In data centers, water consumption refers to the amount of water withdrawn from blue or gray sources. Approximately 80% of the freshwater used by data centers evaporates, while the rest is sent to local water facilities.
Considering that the majority of water used by data centers cannot be reused, monitoring the rate of withdrawal and litres used is necessary. Yet, a 2025 survey conducted on 680 U.S. water stakeholders, including state water managers, found that 54% of respondents did not factor in the rise of technical manufacturer water needs or data centers into their short or long-term planning. The results reveal a serious gap in awareness about the AI sector’s water consumption. In the U.S. AI infrastructure has yet to be federally regulated. In fact, the Trump administration has controversially advocated for deregulation, including a failed provision in the Republican budget bill which would have placed a moratorium on state-level AI legislation for the next ten years. This absence of regulation has allowed corporations like Microsoft and Meta to avoid reporting their water usage in detail, and researchers have found that only 43 percent of data center operators track their water usage.
Monitoring remains weak for two main reasons: there is no common measurement standard, and reporting is rarely mandatory because legislation is rarely enforced. Even if a company provides water usage data, it is difficult to aggregate or compare over time when the information is not standardized across users. It also opens up the possibility of obscuring information through inadequate measurement tools. In Canada, efforts to improve transparency are still limited. Ontario, Prince Edward Island, and Quebec now publicly publish water-withdrawal data, while Alberta and other provinces release only sector-wide totals instead of company-specific usage.
There are major loopholes that undermine data quality and contribute to the problem of unstandardized measurements. In certain contexts, water withdrawal rates are self-reported by organizations and can be based on estimates rather than actual use. Quebec also allows for a 25 percent margin of error in monthly volume reports. Considering that a large data center can use almost 19 million litres of water per day — 589 million litres per month — a 25 percent error margin is a difference of around 147 million litres. That’s enough water to fill 59 Olympic swimming pools. Without transparency, Canada can’t manage water sustainably.
One model for effective monitoring comes from the European Union (EU). In 2024, the EU introduced legislation that mandated data centers to report energy and water consumption, as well as steps they are taking to reduce them. A key aspect of this legislation is that it is EU-wide, which enables cross-country comparison of water usage. This is vital for long-term water management because it allows rates of use to be limited according to predicted droughts or changes in precipitation. Canada and the U.S. should implement similar federal legislation that not only mandates reporting but standardizes it so the data collected can be useful. To prioritize the needs of people over megacorporations, governments can and should hold major companies accountable through tracking.
Canada’s vast freshwater resources have long been taken for granted, but in order to secure a future where clean water is accessible to Canadians, its industrial use must be controlled. Mandatory standardized water-use reporting is an important first step towards accountability. Without it, Canada risks diminishing its most important resource for short-term economic investment. By demanding transparency from the companies that depend on water, Canada can join the EU in setting a global precedent: that technological development should not be adversarial to the longevity of our planet, or the public interest of citizens.
Edited by Gabriela Flaschberger
Disclaimer: This is an article written by a Staff Writer. Catalyst is a student-led platform that fosters engagement with global issues from a learning perspective. The opinions expressed above do not necessarily reflect the views of the publication.
I’m a writer for Catalyst and the McGill Journal of Political Science. I am passionate about topics such as politics, sustainability and geography!