Canada’s New AI Czar And The Energy File
What should be on his Ministerial agenda
In what is certainly a first for Canada, we now have a Minister of Artificial Intelligence and Digital Innovation. Evidently the new Prime Minister did not think it necessary for there to be a Minister of Human Intelligence (probably because that’s a matter for the provinces), but nevertheless, it pleases me that digital innovation is now at the big table in Ottawa. Some good will come from this, along with the risk of some inconvenient rules and compliance reporting.
At next week’s Global Energy Show, I’ll be part of a panel discussion on what should be on the agenda of the new Minister, specifically as the agenda topics relate to energy. Here’s the brief:
The Advent of the Fourth Industrial Revolution.
There’s no doubt that AI is going to dramatically change the way we live and work in every way. And when it comes to energy, AI is both the next big opportunity and, for Canada, potentially the next big customer. AI data centres could change Canada’s energy industry. For the first time in Canada’s Federal Cabinet there is a Minister of the Crown dedicated to AI.
So what should be on his agenda when it comes to energy and AI? What’s the opportunity here? Where are the challenges? Is Canada too late or can we learn from what our allies and our competitors have already done?
Big questions. I have answers.
The Backstory
Canada doesn’t routinely create new ministries, and certainly not in response to specific technologies. There isn’t a minister of Cloud Computing, for example, although perhaps there should be. Blockchain technology, or more precisely, crypto currency, doesn’t have dedicated mandarin oversight, and is split across a few portfolios (Finance, and Innovation and Science). And for all the hype about electric vehicles, another technology, Canada doesn’t have a Minister of Rubber Wheels.
So why AI and digital innovation, and why now?
There are several reasons, I suspect:
There is a general consensus that digital technologies, and AI specifically, will have long-run and deeply transformative impacts on the country and its products, the workforce, the economy, and national security. You only need to watch the latest kick-boxing and dancing robots from China to realize that change is afoot, literally.
The very rapid pace of development and advancement of AI is unsettling. Every week brings a fresh and shocking realization. Yes, AI has been the subject of research and development for decades, but since the release of ChatGPT in December 2023, and DeepSeek this January, previously insurmountable computational hurdles have been summited with increasing ease.
These technologies are, for the present, quite democratic. Anyone anywhere can access them. And for the most part, these technologies in the West are owned by American technology giants (Apple, Alphabet, Microsoft, Meta, Nvidia, Amazon, Tesla, Oracle) and their offspring and siblings. To a degree, much of Canada’s digital future has already been outsourced to a freshly hostile nation, and the alternative (China) isn’t that appealing either.
The capital spend by the digital giants is gobsmacking. The top six spent US$230 billion for data centers and the like in 2024. To put that in perspective, the Canadian Association of Petroleum Producers (CAPP) estimated the total capital spend for the entire Canadian oil and gas industry (Canada’s largest export sector) to be a paltry US$29 billion. The planned Canadian data center investment to 2030 is US$73 billion. Who doesn’t want a piece of that AI action?
Got it? AI is large, transformative, lucrative, moving quickly, dangerous and a major risk.
My favorite analogy about the likely impacts of AI on society is the advent of flight. The Wright Brothers solved the riddles of air travel in 1903, but 35 years had to pass before the first jet engine powered a flight. Fast forward to 2025 and see how far we have come—mass transit by plane, a man on the moon, and ambitions to colonize other planets.
Today’s AI is like jet travel in 1941. We cannot even imagine the impacts it will have.
The Energy Angle
As sexy as AI is, its two biggest inputs are unsexy land and very unsexy energy—land to house the racks of computers and circuits, and energy to run the computers and cool them down. Energy is proving to be the bigger problem to solve:
Data centers demand very high power quality, with reliable flows and no fluctuations or surges, which can damage the computers. Historically, power generation, transmission, and the grid itself have been inherently irregular businesses—most end user gear (fridges, hair dryers, industrial equipment) has been designed with resilience in mind and can tolerate power irregularities. The lights might flicker but they don’t go out.
The amount of energy required for AI is (for now) orders of magnitude greater than for normal computing. A single query on ChatGPT uses 10 times the energy of an equivalent search on Google, and produces 30 times as much CO2 (owing to energy mix, heat load, and compute intensity). Training large language models uses 7-8 times more energy than traditional computing. DeepSeek does highlight that these early stage models are energy hogs, and next generation AI might be way less energy intense.
Clean renewable energy from wind and solar isn’t good enough for data centers because of their intermittent nature. They have to be either coupled with something of greater reliability (gas turbines typically) or with big batteries. Meanwhile, governments everywhere, including Canada, have made energy transition and decarbonization central to their mandates since the Paris Climate Accord of 2015. Even the tech giants themselves have made brand promises of carbon neutrality. The growth of renewable energy has been stellar, but it’s hitting its limits in meeting AI’s needs.
Dispatching electricity from a distance doesn’t work either. Resistance from the wires on which power travels turns into heat, and the longer the wire, the more power is lost. Power utilities thus tend to be local businesses so that power losses aren’t crippling. As a result, big data centers are clustering close to where they can get high quality reliable power.
Those power utilities (and I’ve worked for a fair few) are natural monopolies that move very slowly, at the pace of GDP growth (say, 2% per year). Asking them to grow at pace to keep up with AI energy demand is nigh impossible.
Lessons From Away
The new AI Guy in Ottawa is well advised to pay heed to how other jurisdictions have addressed the challenges of AI and digitalization.
Co-location of Energy and AI.
To solve for the problem of power degradation over long distances, AI operators are locating their data centers as close to reliable sources of power as they can. Energy corridors might be good for residences, but long power lines are critically vulnerable to storms, sabotage, and the occasional squirrel.
AI is Energy.
The technology giants have all learned that their AI strategies are highly vulnerable to energy. Twenty-four months ago they rejected out of hand any energy supply deal that was not pure green—thoroughly brand consistent and terribly woke. Today, they’d incinerate kittens and bunnies if it generated enough power.
More importantly, the energy needed by AI is different from that required by hair dryers and fridges. This creates a new and different demand pull that utilities might struggle to fulfill.
Baseload is Key.
Securing baseload power is the order of the day. The tech giants are aggressively buying long supply deals for nuclear power:
Meta has a new 20 year deal with Constellation Energy for nuclear power in Illinois.
Microsoft has signed a 20-year power purchase agreement to restart the Three Mile Island nuclear plant in Pennsylvania.
Alphabet has partnered with Kairos Power to procure energy from small modular reactors (SMRs).
Amazon has acquired a data center powered by a nuclear plant in Pennsylvania.
Grid Separation.
The tech giants want to distance themselves as much as possible from local utilities and grid suppliers. They need to move at light speed, while monopoly local utilities are governed by regulators, slow permitting processes, and public input.
Staying grid dependent also puts the tech giants in direct competition with their end customers for power. That’s a recipe for power price escalation, and very hard on the brand.
Other jurisdictions are reacting. West Virginia recently passed a law allowing companies to develop their own independent energy grids, to service data center demands. Abilene Texas has also sanctioned a new natural gas plant to be located adjacent to a data center.
The Minister’s Agenda
Here’s some energy suggestions for the new Minister of AI:
On energy diversification. Support energy diversification. Encourage clean energy inputs (solar, tidal, wind) for AI, but only where that energy can meet AI’s demanding reliability needs. Anything less will kill AI.
On energy localization. Work with provincial energy ministries to allow for direct power investments by AI companies and third parties. Embrace power AI co-gen agreements, where some power from an energy development is sold back to the local utility for community use.
On waste energy use. Encourage via the tax code and regulatory reform options to use the waste energy from data centers in agriculture, green houses, and food production.
On AI in energy use. Help create branded Canadian AI products that meet global standards for carbon content to improve their attractiveness to sensitive global markets such as the EU. Change the tax code so that AI-related spend on energy technology with Canadian businesses receives preferential treatment.
On AI and government. Lead by example. Create a federal center of excellence or agency tasked with sharing data center and AI energy best practices, driving code changes for higher power quality, and organizing national summits related to AI and energy use.
On funding for energy AI. Advocate for changes to the tax code to encourage energy AI investments by business, such as a special category of scientific development and research, with emphasis on energy measurement, optimization, carbon tracking, and use.
On AI in the national interest. Unlock federal land close to population centers and available power generation to create AI hubs and centers of excellence. Change land tax and transfer laws to enable fallow land reform specific for AI, data centers, and energy production.
On AI and energy education. Assist with the upgrading of educational curriculums to create a focus on energy use for advanced technologies.
On permitting reform. Encourage permitting reform and uniformity among the provinces so as to encourage and accelerate AI data center and related power developments wherever there is interest.
Conclusion
Canada is blessed with ample undeveloped land and enormous untapped energy resources, which on paper makes the country a fantastic target for investment in AI infrastructure. Even the cold climate is a benefit for an industry that has a heat problem. I’m confident that the new Minister will make a solid go of his new mandate.