The conditions that led to the 2015 Paris agreement have long since been rendered outdated by the advancement of AI and other changes to our world
In 2015 world leaders met in Paris to set the course for climate action and agreed to limit global warming to well below 2°C above pre-industrial levels. Those targets relied heavily on getting to 100% renewable energy, electrifying transport and reducing fossil fuels. But one big factor was left out of those plans: the rapid growth of artificial intelligence (AI) and the massive energy it’s consuming. Now as AI is becoming a pillar of the global economy, climate goals remain stagnant, and we need to ask the big questions about how we reconcile progress with responsibility.
AI’s rapid growth, especially since the introduction of generative AI tools like ChatGPT and MidJourney, has upended industries and created unprecedented demand for computer power. Training and running advanced AI models requires vast amounts of energy, mostly to power the data centers where the computations are done. These facilities use as much electricity as a medium sized city and are straining local grids and making it harder to decarbonize the power system.
The scale of this demand was not factored in when nations were setting their climate strategies in 2015. While many plans accounted for electrification of transport and heating, AI was still an emerging idea. Today the data center industry, driven by AI, cloud computing and internet usage, accounts for about 3% of global electricity consumption and that’s expected to rise sharply as AI adoption grows.
The energy challenges of AI are particularly acute in British Columbia, Canada where a clean electricity grid was once the foundation of the province’s climate strategy. BC Hydro, the publicly owned utility, generates most of its electricity from hydro. But recent data shows BC Hydro can’t meet domestic demand without importing electricity from neighboring regions including Alberta and the US where fossil fuels dominate the energy mix.
In the last fiscal year BC imported over 13,600 gigawatt-hours of electricity – more than double the annual output of the controversial Site C dam, a $16 billion hydro project currently under construction. Importing electricity undermines the province’s green credentials and raises questions about how it will meet future demand as data centers grow to support AI.
Climate goals initially focused on reducing emissions from transport and industrial processes are now being challenged by the energy demands of AI. For example, policies promoting electric vehicles (EVs) assumed electricity demand would grow incrementally but AI is upending those calculations. Data centers designed to power AI workloads require massive energy densities and continuous operation and are adding stress to grids already dealing with EVs and renewable energy integration.
Globally nations are facing similar dilemmas. In the US data centers are driving demand for new natural gas plants even as the federal government is committing to decarbonize the grid by 2035. Meanwhile countries like Ireland and the Netherlands have temporarily halted approvals for new data center connections to protect grid stability and meet emissions reduction targets. These tensions are highlighting the growing challenge of balancing climate goals with the demands of a digital economy which now has the added pressure of AI.
AI and its energy demands have added a new layer of complexity for climate policymakers. Some say the solution is to accelerate the transition to renewable energy and invest in advanced technologies like small modular reactors (SMRs) and energy storage. Others say it’s about improving data center efficiency through liquid cooling and more efficient chips.
But these solutions take time and capital and may not be enough to keep up with the rapid growth of AI driven energy demand. Policymakers will have to make tough choices: should resources be directed towards building more renewable capacity to support AI or should data center growth be limited? And how can we make sure AI’s benefits outweigh its costs?
The AI revolution has blown apart assumptions about energy demand and emissions reduction pathways and we need to face the reality of our existing climate strategies. As British Columbia is trying to balance the promise of AI with a sustainable future the time to act has never been more pressing. A net zero world will require not only innovation but also a willingness to confront the trade-offs that come with plugging in these transformative technologies to our planet.