Could the electricity needed to power artificial intelligence make climate change worse?
The federal government pledged $2.4 billion to build out artificial intelligence projects in Canada in the recent budget on top of billions in federal funding over recent years, and a national strategy on AI.
Canada’s Competition Bureau noted in a discussion paper in March that the AI in Canada is seeing a “rapid evolution” that raises questions about how best to ensure the evolving landscape is competitive and fostering innovation.
At the same time, a report earlier this year from the International Energy Agency warned that “electricity consumption from data centres, artificial intelligence (AI) and the cryptocurrency sector could double by 2026.”
It states the AI industry in 2026 is “expected to have grown exponentially to consume at least 10 times its demand in 2023.”
Alex de Vries, a Ph.D candidate focusing on the environmental impacts of emerging technologies at Dutch university VU Amsterdam, says the electricity AI requires could emit more greenhouse gas emissions.
“The only thing that I can see as a result of this increasing energy demand will typically result in increased use of fossil fuels, which will ultimately boost climate change,” he told Global News.
Data centres, which can house servers used for AI as well as for mining cryptocurrency operations, could consume an estimated 1,000 terawatt-hours (TWh) in 2026, up from 460 TWh in 2022.
Microsoft had said late last year it is spending hundreds of millions of dollars to expand its cloud computing and AI infrastructure in Quebec, which has been in place since 2017 and is home to “fundamental research in artificial intelligence (AI) and machine learning.”
The IEA report states the rise in 2026 power needs are “roughly equivalent to the electricity consumption of Japan.”
The report cites de Vries’ research stating that a request from Microsoft’s OpenAI ChatGPT uses almost 10 times the energy that an average Google search does.
OpenAI’s CEO Sam Altman in January said AI use requires an energy breakthrough.
“It motivates us to go invest more in fusion,” he said at the World Economic Forum, also listing cheaper solar power and storage.
The IEA report states clean energy supply “is forecast to meet all of the world’s demand growth through 2026,” and Microsoft, Amazon Web Services and Google have all invested billions in renewable energy.
But de Vries is less optimistic companies and countries can convert their power grids to renewable energy so quickly.
“There’s a limited supply (of renewable energy),” he said, speaking from Almere in the Netherlands.
“So if we decide to put this (electricity) in data centres, which is a growing segment, then it just means that somewhere, somewhere else is going to have to be powered with fossil fuels.”
He told Global News companies are racing to build better AI models, which requires putting more resources into them, but suggested finding enough power will bottleneck their progress.
He also said he suspects AI will not the be “miracle cure” proponents suggest it might be, comparing it to other technologies that elicited enthusiasm but saw limited use, like blockchain and virtual reality.
Either way, he said countries and companies need to find a way to offset or stop emissions, otherwise the technology of tomorrow will exacerbate a major challenge today.