I step from my car into the morning fog that shrouds the perimeter fence of a massive data center outside Vancouver. The complex sprawls across several acres, its white buildings humming with an energy that powers millions of digital decisions happening every second. The security guard checking my credentials jokes about the weather, but I’m thinking about what’s happening inside—how the artificial intelligence revolution is literally heating up our world.
“We’re running at about 32 megawatts today,” says Elaine Cho, operations director for Pacific Cloud Solutions, as we walk through a corridor flanked by rows of servers stretching into the distance. “That’s enough to power roughly 24,000 homes.” Her casual comparison lands with weight—this single facility consumes as much electricity as a small city.
The explosive growth of artificial intelligence is creating an insatiable appetite for energy. AI systems require vastly more computational resources than traditional computing, with training a single large language model potentially consuming as much electricity as 100 Canadian households use in a year. According to the International Energy Agency, data centers already account for approximately 1-1.5% of global electricity use, and AI’s rapid adoption threatens to push those figures dramatically higher.
When I visited Microsoft’s Quincy data center campus in Washington state last month, the facility was surrounded by new construction—expansion that mirrors what’s happening globally as tech giants race to build infrastructure for AI capabilities. The facility sits near the Columbia River, its location carefully chosen to access the region’s hydroelectric power.
“Companies are facing a fundamental challenge,” explains Dr. Manoj Patel, climate technology researcher at the University of British Columbia. “The computational demands of AI are growing exponentially, but the planet can’t support that trajectory with our current energy systems.”
The numbers support his concern. A 2023 study published in the journal Joule estimates that training GPT-3, a predecessor to ChatGPT, emitted approximately 502 tons of carbon dioxide equivalent—roughly equal to the lifetime emissions of five average American cars. More advanced models require even more processing power.
Walking between server racks generating enough heat to fog my glasses, I think about the communities I’ve reported from across Northern Canada—places already experiencing rapid climate change. In Inuvik, Northwest Territories, elder Sarah Kuptana showed me the shoreline where her family once harvested shellfish, now drastically altered by thawing permafrost. The connection between digital advancement and environmental impact suddenly feels deeply personal.
But this story isn’t simply about problems—it’s about the race to reconcile technological progress with environmental responsibility.
“We’re aiming for carbon-negative operations by 2030,” Cho tells me as we exit into a construction zone where her company is installing heat recovery systems. The recovered thermal energy will warm nearby office buildings and greenhouses, turning waste into resource. “Every kilowatt-hour we save or repurpose matters.”
The tech industry’s awareness of its growing footprint has sparked innovation. Google has pioneered using machine learning to optimize its own data centers, reducing cooling energy by 40%. Microsoft has deployed underwater data centers, using the ocean’s natural cooling properties. Amazon Web Services has committed to powering its operations with 100% renewable energy by 2025.
“The same computational power driving the AI revolution can help solve its energy challenges,” says Amita Singh, climate lead at Digital Future Initiative. “We’re seeing AI optimize energy grids, improve battery technology, and accelerate materials science for better solar cells.”
Beyond corporate initiatives, policy makers are beginning to act. The European Union’s Energy Efficiency Directive now includes specific provisions for data centers, requiring energy efficiency measures and waste heat recovery. In Canada, British Columbia’s CleanBC program offers incentives for energy recovery projects like the one Pacific Cloud Solutions is implementing.
As we finish the tour, Cho shows me the facility’s weather station. “We track environmental conditions constantly—it affects both our cooling needs and our renewable energy generation,” she explains. Behind her, a bank of screens displays energy flow visualizations and carbon intensity metrics in real time.
I’m struck by the duality of what I’ve witnessed—both the environmental challenge of AI’s energy hunger and the technological possibility of its solutions. While visiting communities in Northern British Columbia last year, I spoke with members of the Tahltan First Nation who were negotiating with tech companies about renewable energy projects to power nearby data centers. Their perspective was nuanced—acknowledging both economic opportunity and environmental concern.
“We need to approach this challenge with both urgency and hope,” says Andrew Wilson, executive director of Climate Tech Canada. “The computational power that’s demanding all this energy is the same force that could help us transition to a sustainable grid faster than we’ve previously imagined.”
As I drive away from the data center, its massive cooling towers fading in my rearview mirror, I think about how the digital cloud is anything but ethereal. It exists in physical space, consuming real resources and affecting actual communities. Our digital revolution and climate future are now inextricably linked—each calculation in the cloud leaving an imprint on our warming earth.
The question isn’t whether we continue advancing AI technology, but how we power that advancement responsibly. In the race between AI’s growing energy appetite and our capacity to develop sustainable solutions, the outcome remains unwritten—but the stakes for our climate couldn’t be higher.
