AI can help fight climate change, but what will it cost in power and water?

Embracing new technologies also means asking questions about how they are built and powered, and if they are strengthening our ability to understand and solve problems or making it easier not to.

We tend to think of AI as a mainstream meme generator or deepfake machine. And it is that, but it is also helping drive breakthroughs in medicine, climate science, and energy systems, with real-world benefits and consequences that many people do not see.

For me personally, I used to spend hours buried in government reports, academic papers, and energy datasets, squinting at charts and manually cross-checking information from one source to another. Now, with Artificial Intelligence (AI) tools, I can fact-check climate claims and find relevant studies in a fraction of the time.

It is like having a research assistant who never needs a coffee break, does not complain about your messy workspace or pyjamas, and occasionally reminds me that yes, wind energy really could work off the Great Lakes, or that making Vaughan, Brampton, and Downsview larger transit hubs instead of building the proposed Highway 413 could reduce emissions and congestion while improving connectivity. The information is there if you know where and how to look.

The ability to find out what is possible, plus the pace of discovery, coupled with the increasing daily demand for these tools, makes me realize just how powerful AI is, and not just for climate science, advocacy, keeping decision-makers accountable, and helping the public navigate a world full of misinformation.

At the same time, as AI becomes better at finding information, summarizing reports, generating ideas, and answering questions, it raises another question: are these tools strengthening our understanding, or are we beginning to rely on them for thinking we once did ourselves?

Companies are investing billions of dollars into systems that can analyze huge amounts of data, identify patterns, and generate insights at speeds humans alone cannot match. A 2025 study by the Grantham Research Institute found that AI applications across electricity, transportation, and food systems could reduce global greenhouse gas emissions by 3.2 to 5.4 billion tonnes of CO₂ equivalent per year by 2035 through better management of electricity grids, cleaner energy systems, and more accurate predictions of future energy needs. That is roughly the same as taking 700 million to 1.2 billion cars off the road. These potential reductions could outweigh AI's own energy use, showing its potential as a climate tool.

In practice, AI is already improving wind-power forecasting and helping predict extreme weather. It is also helping reduce reliance on fossil-fuel backup systems. In medicine, it has sped up disease research and helped identify promising new antibiotics to fight drug-resistant bacteria. These advances point to real potential for both climate action and human well-being.

Yet these benefits come with growing demands for energy and water. In Ontario, the reported proposed 720 MW data-centre development in Milton shows the scale of the resources AI infrastructure may require. If built, it could use as much electricity as a large city and require millions of litres of water each day for cooling.

Globally, AI infrastructure could require billions to trillions of litres of water every year. New technologies are helping improve efficiency, including machine-learning systems that optimize cooling. However, the overall water footprint extends beyond the walls of a data centre through the electricity systems needed to power it.

This challenge is not unique to AI. Ontario is also moving forward with gravel and stone extraction, critical mineral mining, population growth, housing construction, and major nuclear-energy expansion.

Recent legislative and regulatory changes have focused on speeding up approvals and streamlining development. While these changes may help projects move ahead more quickly, they have also reduced some of the checks and balances around who is responsible for assessing the combined impact of growing water demand on underground water supplies, watersheds, and drinking-water sources.

Water experts have long warned that groundwater does not always replenish as quickly as it is used, especially when gravel extraction, urban growth, and land-use changes interfere with nature's ability to filter and store water. Groundwater can only provide so much. It does not follow project boundaries, ministry mandates, or industry sectors. It does not distinguish between mines, municipalities, highways, AI data centres, or energy projects. All of them draw from the same shared resource.

Climate change adds another layer of pressure. Ontario as a whole is unlikely to run out of freshwater, but a warming climate is expected to bring greater swings in water availability, more frequent droughts, higher evaporation rates, and added stress on groundwater systems. The question is no longer whether individual projects can be supported today.

The question is whether Ontario's water systems will remain resilient under the combined pressures of development and climate change, and whether some areas, including Waterloo Region and parts of southwestern Ontario that are already facing challenges, could put their future water security at risk.

At both the federal and provincial levels, AI oversight, including Canada's new "AI for All" national artificial intelligence strategy, remains focused on how AI is used rather than on the infrastructure that supports it or the impact on local communities. "AI for All" and Ontario's emerging frameworks emphasize transparency, accountability, bias prevention, and human oversight. These safeguards are important. However, they leave a gap in how large-scale AI centres are considered overall.

In practice, safeguards are often applied one project at a time, while the bigger picture of combined impacts on water and energy systems receives far less attention.

This broader question is beginning to emerge across Canada. In Manitoba, Premier Wab Kinew recently questioned whether large-scale AI infrastructure that requires significant new energy generation delivers enough public benefit to justify its environmental footprint. Regardless of where one stands on that debate, it points to a larger issue that goes far beyond any single project. AI has the potential to deliver tremendous benefits, from medical breakthroughs to climate solutions. But if meeting growing demand requires greater reliance on fossil-fuel generation, those gains could be undermined.

The challenge is not only managing the impacts of individual projects, but also understanding how growing demands from AI, housing, transportation, industry, mining, and energy production interact within the same limited water and electricity systems. The goal should not be to slow innovation. It should be to ensure AI is powered by clean energy and developed in ways that provide broad public benefit while remaining consistent with long-term climate and sustainability goals.

When I asked AI what the solution is to find compromise between those who support it and those who oppose, this was the response for humans to ponder: What level of AI development delivers meaningful public benefit while remaining within environmental limits, supporting climate goals, protecting water resources, and preserving the human skills that technology is meant to enhance rather than replace.

As I finish writing this piece, the irony isn’t lost on me. I used AI to fact-check my research for this column while reflecting on the environmental footprint of the very tools I have come to rely on to save valuable time. Even as I have made considerable efforts to reduce my personal emissions, the convenience and speed of AI come with their own costs.

It is a small but important reminder that embracing powerful new technologies also means asking hard questions about how they are built, how they are powered, and whether they are strengthening our ability to understand and solve problems or simply making it easier not to. Balancing the benefits with the impacts remains a responsibility that belongs to all of us.

Read this article on Newmarket Today.