A question looms for the Great Lakes: with our water protected from diversion to far-flung locales, are we equally ready for an influx of big new water users here at home?
The Great Lakes hold the world’s largest supply of surface freshwater. It is a truly massive amount of water – but also a finite resource that must be managed responsibly for today and tomorrow. The 8 Great Lakes states and 2 Canadian provinces wisely recognized this when they agreed to the Great Lakes-St. Lawrence River Basin Water Resources Compact (Compact) and provincial Agreement. The Compact and Agreement do four things:
- Prohibit diversions of Great Lakes water (with limited exceptions);
- Require the states to manage their own Great Lakes water use within the Basin;
- Set water conservation and efficiency goals and objectives; and
- Establish common water use reporting protocols.
The Compact’s prohibition on diversions ensures that Great Lakes water stays in the Great Lakes Basin. Proposals to pipe Great Lakes water for any use to a location far outside the Great Lakes region are legally not in the cards – the Compact prohibits it. Any exception to the diversion prohibition can happen only in a county that is partially in the Basin, and the water must be returned after use. Beyond the legal constraints, moving vast amounts of Great Lakes water away from our region makes no economic or logistic sense. The good news is this means that Great Lakes water will be staying in the Great Lakes.
But our abundance also makes the Great Lakes region attractive to industries that require large amounts of water.
Rapid Growth Driving Unprecedented Demand
Historically, this meant steel production, petroleum refining, agriculture, and power generation. Today’s growth industries look somewhat different. To be clear, all large-scale water use has the potential to impact our water supply. However, the rapid growth of digital consumption of generative artificial intelligence (gen AI), cloud computing, internet of things, and digital services is driving an unprecedented demand for new data centers. Growth projections vary and depend on a variety of factors, but we can make a few generalizations about the trends in this region:
- Demand for data centers is rapidly increasing.
- Gen AI is a key driver of the growth, requiring vast data processing capabilities.
- The type of hyperscale data centers that can handle big tasks require both large quantities of electricity and water for cooling.
Hyperscale data centers can take up over 10,000 square feet of floor space and house over 5,000 servers that demand water and electricity 24 hours a day. How much water a data center needs varies depending on the size and type of cooling technology used, but hyperscale data centers can use between 1 and 5 million gallons of water per day (MGD) when evaporative cooling, currently the most common method, is the method used.
To put this in perspective, a hyperscale data center that uses 365 million gallons in a year (or 1 MGD) is equivalent to what roughly 12,000 Americans use in a year.
How Do Data Centers Use Water?
Most of the water used in evaporative cooling is used consumptively – meaning the water is not returned to the watershed, but rather, as the name implies, lost to evaporation. The increased pace of this consumption is also alarming. A new report shows that in 2023, U.S. data centers directly consumed about 17.4 billion gallons of water, and the authors expect that figure to double by 2028. Yet only 1% of the water in the Great Lakes is renewed each year, underscoring the hard choices and complex planning that goes into keeping sustainable amounts of water available.
Other cooling methods such as liquid immersion or direct-to-chip cooling consume less water and electricity by directly using water to cool equipment but can introduce contaminants in the process. These methods may be more efficient from a water consumption standpoint but will pose water quality questions. Air cooling is another alternative but uses more electricity. Finally, the reuse of non-potable water and recirculation have the potential to offset consumptive use, but these technologies are not contemplated by most states’ laws and therefore take more time and effort to finalize partnerships and to permit. There also isn’t good data to show how many data centers are adopting them.
Generating electricity to meet data centers’ needs via coal, natural gas, or nuclear fired power plants also requires water. Some refer to this relationship between water and energy generation as part of the “water-energy nexus.” Per the Great Lakes Regional Water Use Database, 70% of Great Lakes reported water use in 2023 was associated with generating electrical power. That’s overall water use (not the percentage of consumptive use), and that percentage generally matches each state’s water use as well. Because the electric utility is the entity that reports its water use and corresponding consumptive use to the Database, we don’t have a clear understanding of what the total water footprint of an individual data center or the data center industry is.
New research also shows that it’s taken just four years for the total capacity of hyperscale data centers (megawatts of load a data center can handle) to double, while both the number of facilities and average capacity rapidly climb. To meet the demand on existing power grids, states will have to add capacity while also meeting state renewable energy targets. In some states, that may necessitate reactivating or expanding nuclear power plants. That’s already happening in Michigan (Palisades), New York (Three Mile Island), and Ontario (Bruce). In others like Ohio, it’s driving the construction of new natural gas plants. The corresponding increase in the cumulative use of water by data centers – both in their indirect energy needs and direct cooling needs must be better quantified and understood.
Transparency is Key
Data centers should be transparent about their total water footprint from the early stages of proposed development. They should also have systems in place to accurately measure water use. It’s estimated that less than one-third of data centers are currently tracking water usage. In the Great Lakes states, when a large water user obtains its water through a municipal water system that has the capacity to supply it, the obligation to track and report water usage rests with that system – not the water user.
Without information about what a data center proposes to use up front and reporting to determine how much water is being used, it’s not possible to fully understand and assess the impact of an individual data center or any large water using industry on a water resource. Better accounting and reporting requirements are needed to guide decision making and protect water resources.
At the same time, state legislatures, economic development agencies, and local governments are inviting data centers to locate in the Great Lakes region with tax incentives and other benefits packages. But when decisions are being made about where to incentivize development, water doesn’t appear to be holistically factored into the equation. That can and should change as states (like Ohio and Indiana) are undertaking studies to better understand water demand and capacity. If not, states may max out an area with data centers and not have the capacity to handle any other type of growth or economic development. Without careful planning, it could also have unintended consequences like depleting groundwater availability in existing private drinking water wells and wells used for agricultural irrigation.
A data center can employ as few as 10 to more than 100 people, depending on its size – but supports 6.5 jobs for every one job directly employed. If data centers turn out to be the economic development engine they have been touted as, then that may drive population growth, further increasing the demand for both water and electricity. For example, the Columbus, Ohio metropolitan area, where data centers are being constructed at a very rapid rate, was the fastest growing city in the U.S. for the second half of 2023.
Climate Change is Scrambling Assumptions
Climate change impacts on surface and groundwater supplies compounds these issues and underscores the need for detailed planning and informed resource management. State laws and regulations are currently not designed to proactively manage water resources in anticipation of how climate change will reshape surface and groundwater flows, large scale water uses, and population patterns. This is especially true for groundwater. While the majestic surface waters of the Great Lakes are prominent in the minds of many, between 40-75% of Great Lakes state residents get their drinking water from groundwater – much of it connected to the Great Lakes.
States should be examining all large-scale water uses of groundwater, including for data centers, to determine whether they’re appropriate in a given watershed or basin and whether those uses pose enough of a threat to other nearby watersheds that the use should be avoided altogether. This examination should include assessment of not only whether supply can meet demand, but what impact the use will have environmentally and ecologically on the resource.
As climate change continues to dramatically alter precipitation patterns, laws will also need to change and adapt in concert to ensure there’s enough water to support economic development and protect our most precious shared freshwater resource, the Great Lakes.