CDN NEWS | 
US NEWS 
Data centres use far less power than EVs or heavy industry, and Canada’s hydro grids and natural gas make it ideal for hosting them.
By Geoff Russ
By Resource Works
More News and Views From Resource Works Here
It was inevitable that artificial intelligence would become the next villain in the global debates about electricity and climate impacts. Examining the numbers shows that the fears about AI, “devouring” power, as some contributors to the Toronto Star have fretted about, are truly unwarranted in the context of industrial energy usage worldwide.
The numbers of the International Energy Agency (IEA) find that all data centres worldwide used about 415 terawatt-hours of electricity in 2024. That is roughly 1.5 percent of global demand, with consumption expected to rise to around 945 terawatt-hours by 2030, which is under three percent of the total.
A recent IEA analysis found that the extra 530 terawatt-hours of data-centre demand to 2030 would account for only about eight percent of global electricity-demand growth. This number is far less than expected from the growth from electric vehicles, air conditioning or heavy industry in general.
Putting consumption in context
AI workloads are just a small slice of that. AI has driven about five to fifteen percent of data-centre power use, with the rest going to other tasks like online searching, video, cloud storage and everyday digital services. Will those be shut down too?
Other sectors of the economy suck up far more power. In 2024, the world’s electric vehicles used about 180 terawatt-hours of electricity, or about 0.7 percent of global energy consumption. Bitcoin mining alone consumes roughly 120 terawatt-hours per year, accounting for up to nearly one percent of global electricity use.
Established industries like aluminium smelting industry’s smelters, long run with cheap hydro dams, are responsible for about four percent of global power consumption.
Against that backdrop, it is silly to treat AI as a uniquely intolerable new industry on the basis of energy consumption. The world’s demand for electricity grew by more than 1,200 terawatt-hours in 2024, with over 80 percent of that increase met by renewables and nuclear power production.
The main modern drivers of energy consumption are
Canada’s clean energy advantage
In that “age of electricity”, the main drivers of consumption are the electrification of transport, buildings and industry, not data centres in isolation. For Canada, the situation regarding AI is very favourable.
The Canada Energy Regulator notes lists more than half of Canadian electricity as coming from hydro, with provinces such as Quebec, British Columbia, Manitoba and Newfoundland and Labrador routinely generating over 85 percent of their power from water. On the whole, more than 80 percent of Canada’s grid is already non-emitting.
The growth of AI data centres is manageable. A recent federal market snapshot counts roughly 239 data centres operating in Canada, in our low-cost hydro regime and cool climate make it, which is attractive for low-carbon computing. Hydro-Québec’s latest supply plan anticipates about 4.1 terawatt-hours of additional demand from data centres between 2023 and 2032 That is growth in the low single digits. Ontario’s Independent Electricity System Operator expects the province’s total electricity use to rise by 75 percent by 2050, driven “in large part” by broad electrification and industrial growth. Data centres are just one of these among several major loads.
Avoiding the coal trap
If Canada does not host AI workloads, it will go elsewhere, perhaps to jurisdictions that are still packed with coal-heavy grids. Relocating AI production will only shift the industry away from hydro-based systems in British Columbia and Quebec, provinces that have ready-made supplies of natural gas as backups, if needed, not heavy oil or coal.
As an exporter of electricity that wishes to help lead an energy transition, spurning such a high-value digital industry like AI on the grounds of climate action is a strange way to reduce global emissions.
We can, and should, still exercise diligence and caution when it comes to energy usage, but the evidence points towards electrification only increasing. Solar, wind, hydro, and nuclear power are already doing the heavy lifting for clean energy generation, and Canada has plenty of room to use its natural gas reserves to cover any shortfalls.
It is not rational to single out AI as a pariah in the world of energy and emissions. AI is just one more industrial customer in an electrifying world, not the great threat to Canadian climate goals that some would have us believe.
Share This:
