Data Centres Are on Track to Wreck the Planet. Can We Stop Them?

Right now, right this second, there’s a huge warehouse somewhere with row upon row of servers humming away as electricity pours into them. Below that hum, a softer one: the sound of water coursing through pipes and running directly over chips that are creating vast amounts of heat. Deep in the caverns of this centre, wherever this place may be—an industrial park in Finland, or next to a suburb in Ajax, Ontario, or a massive tech compound in Texas—along the endless rows of metal and plastic and thousands of kilometres of wiring and circuit boards, there is a tiny, blinking light on a faceplate. In that framework are trillions of “1”s and “0”s spinning through an eternal sea of information. That small, blinking light in those swirling bits of information—that is you, dear reader. That is you, right now, reading these words.

• Hyperscaled data centres require upward of 100 megawatts of power and can use 5 million gallons of water per day
• While some centres have implemented green energy technologies, plenty rely on fossil fuel power
• AI’s harsh environmental footprint is something we must think critically about when engaging with this technology

Most of us use our phones, check our email, do some banking, stream movies and shows, and we have a remote awareness at the back of our minds that we are engaging with a server somewhere. We have a far less clear understanding of how data centres enable our daily lives: the electrical grid, the water systems in our homes, the production and transportation of our food, the police forces, the hospitals, the entire cities and towns we live in.

Across Canada and around the world, enormous data centres are popping up in the midst of suburban developments and rural topography alike. These giant buildings are what feed the digital appetites of our modern society: long steel racks with servers stacked up to the ceiling and trillions of bytes of data being processed every second, information running through fibre optic cables across towns, countries, and oceans. And now, with the explosion of artificial intelligence and our ever-deepening reliance on it, data centres have become a major part of our modern life as well as a key concern in how we manage our relationship with the world around us. Consider this: due to the greater computational power required by the AI-processing architecture, the energy draw of these facilities doubled just between 2017 and 2023.

But what are data centres? How and where are they built, and how do they impact our lives, both digital and physical? What are the benefits and the tripwires associated with their existence, and will this very rapid expansion be a footnote in the history of the future or a cautionary tale about development without limits or meaningful regulation?

The process of selecting a site for the construction of a data centre is complex and involves many factors, but a few make or break the decision.

First, access to power and a reliable electrical grid. This is only a start, as even large cities struggle to meet the appetites of these facilities. Next, the availability of existing fibre optic lines and systems is crucial. If a new centre needs to build a kilometres-long network, the cost and time involved increase incrementally with each metre. A robust local network allows a new centre to tap into what is already there, with future expansion, an inevitability in this space, better aligned within existing corridors.

Water access remains one of the most critical and contentious aspects of development. Water is used for cooling in a myriad of different ways, and systems rely on the existing supply infrastructure and nearby natural sources.

Land prices and availability are part of this stack of decisions, and with what now seems to be a race to construct and expand, some developers are looking beyond existing industrial parks and tech campuses. Some are trading easy access to large networks for access to rural landscapes, especially in farm country. One reason for this is zoning, and many players have found governments in smaller communities eager to work with groups building data centres.

There are numerous legal and regulatory hurdles, and some communities have endeavoured to streamline these bumps. In Canada, many municipalities in British Columbia, Ontario, and Quebec have implemented programs uniquely for this. And the Alberta government has created a strategy to specifically attract AI data centres based on three pillars: power capacity, sustainable cooling, and economic growth. Indeed, Alberta has been trying to establish itself as a hub for future data centres, announcing a $100 billion investment and several planned builds which would put it front and centre on the world map.

Canada has several large contractors who are experienced in building facilities such as these. Construction companies like Bird, PCL, and others combine their talents with data centre developers and operators, like STACK Infrastructure and Cologix respectively, to name just a few, and create the buildings and systems together. They have become so adept that these projects are going up in times that would have been inconceivable even a decade ago.

The big worldwide players, such as Amazon Web Services, Microsoft, Google, and Meta, are sometimes referred to as self-build hyperscalers. They build their own facilities or utilize local talent under a heavy cloak of confidentiality and security, setting up whole campuses.

Data centres vary vastly in scale. Some can exist in warehouses, but the real enterprise today is in giant, hyperscaled ones: centres using over 5,000 servers, in an area covering thousands of square feet. These have become necessary in the age of AI and the dramatic uptick in processing and cooling that it requires. Canada is now home to over 300 data centres, about thirty of these being hyperscaled, with the US leading worldwide with a staggering 5,400—more than in all of Europe combined. The data centres in Canada are concentrated around the major hubs of Toronto and Montreal, which house roughly seventy and fifty respectively. Others are found around Vancouver, Calgary, and Edmonton, and most major towns have one or two in close proximity.

Large industrial structures require deep foundations to provide structural stability for the complex systems they will contain. Geotechnical drilling is conducted to establish the conditions: Where is bedrock? Where is the water table? Will this ground be able to support the structure? Engineers design a foundation treatment to decide what sorts of solutions will be needed. Where the ground is soft, like in Richmond, BC, mass excavations may be carried out to remove the soil, or steel piles may be hammered deep into the ground for the building’s footings to stand on.

However the ground is prepared, tons of concrete is brought in next by mixer trucks, from a facility that must be nearby as the wet concrete has a short lifespan before it must be rejected. The concrete is poured into forms—wooden bracing which holds it in place until it has cured enough to take shape. Steel frameworks are built upon the foundation structures, followed by the more complex work: the installation of hardware, the endless rows of servers and support structures, the kilometres of fibre optic cable, the military-grade security systems.

As more of these massive, complex structures take over our landscapes, we have to ask ourselves: What implications do they have for our everyday lives, specifically in their demands on our precious supplies of power and water?

Standard data centres generally require five to ten megawatts of power. By contrast, hyperscaled ones can draw upward of 100 megawatts—enough to power a small city. If a data centre used 100 megawatt hours (MWh) over the course of a year at 100 percent capacity round the clock, it would consume 876,000 MWh of electricity, though actual use ranges closer to the 80 percent mark. Can municipalities in Canada meet those sorts of energy needs while providing the electricity that society at large uses?

I looked at some of the places where the future is already unfolding. Northern Virginia is home to the greatest concentration of large data centres in the world. The reasons for this are numerous—proximity to some of the most important facilities and organizations on the planet being high on that list. Around 70 percent of the world’s data passes through this corridor. What this has meant for local consumers of electricity is a surcharge in their power bills and the potential for further price spikes once the current rates expire. This has been attributed to the power demands of the data centres.

In my home province of New Brunswick, a proposed data centre is expected to require 380 megawatts of power when it reaches completion. To put that in context, it is nearly half the amount of power to be produced by an upcoming gas plant in Tantramar, New Brunswick—a plant that was proposed because the province is already falling short and envisions difficulty providing power for future population growth.

Data centres also rely heavily on supplemental power to cover gaps or potential outages while they run. Google, for instance, is considering the use of a small-scale nuclear power plant to make up its deficits and also to move away from fossil fuel–dependent power generation, with a demonstration reactor planned in Tennessee.

Or take the case of Caterpillar—a company known for producing large heavy equipment such as dozers, excavators, and mining-haul trucks—which saw a remarkable uptick in its share value when the firm became the go-to supplier for something else: industrial-scale diesel generators. Which is to say that not only do the centres require massive amounts of power but the prevailing understanding often is that fossil fuels will fill the gaps.

This is problematic because data centres have long been advertising their green credentials. Indeed, they were some of the early adopters of alternative sources, at least in their public messaging. A defining feature of many centres is their windmills and solar panels, and data centre developers have been calling for greater investment in these technologies. The reality, however, is more complicated. For instance, Alberta’s primary draw for developers is its easy access to gas for power generation plants. The province has leaned into this appeal, skipping the window dressing of alternatives, and a Swiss company is the latest to jump in, with a plan for an $12.6 billion data centre.

At a time when we need to drastically move away from the use of fossil fuels, how have we brought ourselves to a point where we are embarking on an expansion of their use?

Then there’s water. Hyperscaled data centres can consume around 5 million gallons per day, equivalent to the water use of a town of 10,000 to 50,000 people. That is 19 million litres. Some estimates have stated that a 100-word AI answer consumes the equivalent of one bottle of water. Some water is discharged while some is lost to evaporation through the cooling process. The discussion around water also depends on how we quantify its overall use. Are we talking just about what is used directly in the processing? Or should we also include what’s used in the construction of the facility itself—what goes into concrete, steel production, other materials and consumables, and power generation.

Cooling architecture varies in relation to what it is cooling. Some facilities use a hot/cold aisle containment, where cool air is piped in from under the floor and cycled across the servers, and hot air is extracted and blown out the other direction. More advanced systems—used for newer technologies like chips created by companies such as Nvidia which specialize in high-power AI uses—incorporate direct liquid cooling, where water or another fluid is piped across the chip architecture to cool it.

The cooling in data centres is conducted in different ways depending on the needs of the facility and the environment it is in. During the explosion of crypto processing for things like bitcoin, cold regions, such as Iceland and other Nordic states, found themselves at the centre of development, with facilities using the long and cold months to take advantage of access to the freezing cold air and water for cheaper and easier cooling. It is here, in the Nordics, that we can see some creative and thoughtful planning. In southern Finland, the town of Mäntsälä has been channelling heat from a nearby data centre to homes and water systems. This is but one way in which some benefit can be derived from what we may only see as a problem. Surely, Canadian cities, universities, and companies can be leaders in developing these sorts of technological solutions.

Regardless of the cooling technology, it remains dependent on water supply. Data centres draw on municipal supplies and, perhaps more worryingly, also directly from the local water table, through wells which they drill and pump from. This is at the crux of many disputes regarding water use and ownership. Cases of municipalities having far reduced water pressure and access soon after data centres came up have made headlines across North America. What happens when rural communities have to choose between watering crops and providing water to data centres?

• After Intense Lobbying, Carney Allows Gas-Powered Data Centres in Alberta
• Big Tech Is Hiding the Environmental Cost of Chatbots
• Evan Solomon Wants Canada to Trust AI. Can We Trust Evan Solomon?

At this point in 2026, we would be irresponsible not to learn from dramatic failures in water management elsewhere. Iran is in the midst of an existential threat to its regime, but they are also faced with a collapse of another kind: Tehran is out of water, with reports that taps are running dry after alarm bells rang for months. Decades of profound corruption, mismanagement, and ignoring calls from experts (indeed, often imprisoning those experts) led to critically low levels in aquifers and storage bodies. These bodies are reported to be so far past the point of no return that even heavy rains will not repair the damage. And Iran is not alone.

Across the globe, water access and scarcity are sparking conflicts and mass migration. We have seen access to water being used as a weapon of war against the most defenceless populations. Closer to home, New Brunswick has been in the grips of a once-in-a-generation drought. We in the West are not beyond the reach of such catastrophes, and how we manage these resources now will determine how we weather future storms.

What’s worse, the centres themselves are opaque about how much water they are using. In fact, in the US, companies have sued to prevent this information from being released to the public or even claimed that they were not quantifying how much they used and so did not have information to share. After having worked on large construction projects around the world for over thirty years, I can confidently say these companies know how much water they consume—down to the litre.

That may make you angry, but then do we, as citizens, have a concrete grasp of where we fit into all of this? To be clear, data centres are only one part of our society that consumes water and electricity. Food production, mining, oil extraction, manufacturing of clothes—all these have a massive water footprint (you may be dazzled by the water footprint of a cheap cotton T-shirt). It is relatively easy to comprehend our household water use. But we have a much poorer understanding of what activities and processes consume water and power out of sight and on our behalf.

And that being the case, we have to ask ourselves if new technologies like AI are being utilized in the best way. There’s a profound difference in use cases. Using AI and other complex information systems to advance research and development on a thousand fronts, from medicine to technology to energy and transportation, would be a worthwhile endeavour and what these technologies were supposedly developed for in the first place. But an AI video of your cat dressed as an astronaut? Or pretending to create art and literature while avoiding the effort, learning, discipline, and time that actual creation demands? These hardly justify the use of scarce resources.

When we better understand the AI footprint, as well as the impacts of our choices, we get a clearer lens through which to see how we use the technology. Data centres are going to consume vast amounts of water and power, but we need to make better decisions around how, when, and why.