Whether you embrace it or not, artificial intelligence (AI) is increasingly integrating into our daily routines. Be it browsing online stores or using search engines, AI is rapidly turning into an invaluable, efficiency-boosting asset for both individuals and businesses.
When it comes to
climate change
AI is also demonstrating its value in these areas. At the United Kingdom’s Cambridge University, scientists are employing AI across various fields such as climate modeling and land-use planning, viewing it as a revolutionary instrument for safeguarding nature.
Scientists from Oxford University have developed an artificial intelligence tool aimed at enhancing transparency regarding corporate environmental practices. Notably, even tech giant Google has endorsed these advantages, working on multiple AI-driven solutions to boost climate resilience.
Regardless of AI’s considerable potential to bring about beneficial effects
climate crisis
There are worries about its substantial potential to increase greenhouse gas emissions.
According to a recent report by the International Energy Agency (IEA), artificial intelligence is causing a significant rise in electricity consumption.
Data centres
, which serve as the foundation for AI systems, are expected to see their energy consumption double over the next half-decade.
The IEA forecasts that by 2030, the power consumed by data centers will increase to approximately 945 terawatt-hours, which exceeds the total electrical usage of Japan.
Nevertheless, the report also highlights that AI could potentially reduce
emissions
elsewhere. It says that if it is adopted in the right ways, the carbon savings it accounts for could offset the additional emissions it generates.
AI is one of the biggest stories in the energy world today,” says IEA Executive Director Fatih Birol. “But until now, policy makers and markets lacked the tools to fully understand the wide-ranging impacts.”
How much energy does artificial intelligence require?
Artificial intelligence demands substantial energy for both training and operation. The significant computational resources needed to sustain extensive language models derive from numerous servers located within data centers.
data centres
, with some consuming as much electricity as an entire smaller nation.
Data centers can be found across the globe, but the United States stands out with 5,381 installations, accounting for approximately 40% of the worldwide market. Significant numbers of data centers also exist in other nations such as the United Kingdom, Germany, India, Australia, France, and the Netherlands.
The energy usage of these establishments is considerable. Some
AI-focused data centres
Consume as much electricity as two million homes. In 2023, these entities made up about 1.5 percent of worldwide electricity usage, yet their demand is expected to significantly increase in the following years.
Training AI requires a great deal of processor power, and therefore a lot of electricity. Research published in the Journal of Machine Learning found that training the popular OpenAI
ChatGPT
The model utilized 1,287 megawatt-hours of electricity, equivalent to the amount of CO2 produced by 80 short-haul flights within Europe.
Noman Bashir, who holds the position of Computing and Climate Impact Fellow at MIT, explains, “The distinctive aspect of generative AI lies in the high level of computational power it demands.” He further elaborates, “At its core, it’s merely another form of computation; however, a single generative AI training cluster can use up to seven or even eight times as much energy compared to an average computing task.”
Operating the software consumes less energy for each individual task; however, this can rapidly accumulate in terms of overall power usage when millions of inquiries are processed daily.
The Electric Power Research Institute discovered that, for each inquiry,
ChatGPT
uses roughly 2.9 watt-hours of energy, which is about ten times more than what is needed for a typical Google search.
By early 2025, ChatGPT is handling over a billion daily inquiries, with this number continuing to increase.
In early 2025, approximately 8 percent of American adults were utilizing ChatGPT.
primary search engine
That figure remains significantly lower than those using Google, yet considering it has surged from merely 1 percent since June 2024, this highlights the swift transition toward AI-driven utilities.
Additionally, we must take into account the evolving nature of AI. Present inquiries typically restrict themselves to text-only engagements. As new developments emerge,
AI video
, visual and auditory applications have no precedents, but they are expected to be even more ravenous for resources.
When considering the environmental effects of generative AI, it’s not solely about the power used when connecting your computer,” explains Elsa A. Olivetti, who is a professor in the Department of Materials Science at MIT. “The repercussions extend far beyond that to systemic levels and continue depending on the choices we make.
Is it truly possible for AI to compensate for its own carbon footprint?
As per the IEA, worries that
Artificial Intelligence might speed up global warming.
The report suggests that ‘overstated.’ Even with the increase, emissions from data centers will remain a small portion of global energy-related emissions, approximately 1.5 percent.
The argument posits that broad acceptance of artificial intelligence might enhance numerous operations’ efficiency, thereby decreasing emissions in various sectors. This potential reduction can stem from improved industrial procedures, enhanced scientific exploration, or breakthroughs in technological advancements.
The IEA suggests that widespread use of current technologies could
AI-led solutions
could lead to emissions reductions of up to 5 per cent by 2035. It claims this will offset the increase in emissions generated by data centre demand.
A separate report from Energy Intelligence predicted a doubling of energy demand, but also framed AI as a key enabler of the clean energy transition.
It mentioned improved grid management, decreased costs in low-carbon technologies, and better incorporation of renewable energy sources as key factors.
benefits AI could bring
It was also suggested that improvements in processor efficiency, advancements in cooling technologies, and optimizations in algorithms will eventually reduce AI’s significant energy consumption.
Even though the IEA report views the future of AI and its climate effects positively, it points out that this result is not guaranteed.
The report emphasizes that “there is presently no significant drive to guarantee broad acceptance of these AI applications.” Consequently, it adds, “their overall effect might remain minimal by 2035 unless the required supportive environments are established.”
Harnessing the full potential of AI necessitates focused efforts across various sectors. Notably, it highlights how AI can significantly benefit the energy sector by enhancing grid management and distribution—a field where AI remains largely underutilized today.
It also admits that investment in low-carbon electricity generation is crucial, particularly when it comes to supplying
energy-hungry data centres
.
Certain players are advancing in this area. Amazon stands as the biggest corporate purchaser of renewable energy globally. The company asserts that more than 90 percent of its activities, encompassing its Amazon Web Services data centers, are now fueled by renewable sources.
With more than 300 data centers globally, Digital Reality has pledged to use renewable energy sources. Currently, all the energy requirements for their European properties are covered by purchased renewable energy.
However, transitioning to environmentally friendly practices isn’t simple for the data center industry. The inconsistent supply of renewable energies poses a hurdle, along with geographic constraints that can affect access to sustainable power options.
The part played by the United States in shaping AI’s influence on future climatic changes.
With most of the world’s biggest data centres in the US, this will be where the largest growth in energy demand will be seen.
By the end of the decade, energy consumption from data centres is projected to outstrip that of all other energy-intensive activities combined (production of aluminium, concrete, chemicals, etc.), according to the IEA report.
Currently, U.S. data centers depend primarily on fossil fuels, notably natural gas. The IEA does not anticipate this trend shifting, especially considering the present administration’s emphasis on utilizing these conventional energy sources.
Just this week, President
Donald Trump
issued an executive directive asking cabinet officials to pinpoint areas where coal-based facilities could assist in hosting AI data centers.
In Louisiana, plans have been initiated for building a substantial natural-gas-powered facility aimed at supporting the operation of an enormous new data center currently under construction.
Meta
.
The IEA’s report outlines a scenario that can only be realized through coordinated actions and strong political backing. According to prevailing priorities, AI might equally well be employed to discover new oil and gas deposits as it would be to identify methane emissions or enhance grid efficiency.
When discussing how AI might ‘counterbalance’ its emissions, we must consider the broader implications. Since carbon dioxide remains in the atmosphere for centuries, even if AI ultimately discovers methods to reduce more emissions than it generates, it cannot undo the harm caused during the interim period.
The report states that “the broad implementation of current AI technologies might result in emission cuts significantly greater than those produced by data centers; however, these reductions would still fall short of what is required to effectively tackle climate change.”
Leave a Comment