The climate trap of artificial intelligence

Artificial intelligence is becoming one of the defining pressures on the global energy system, creating growing tension between climate progress and the enormous electricity demands of the digital economy. According to energy think tank Ember’s Global Electricity Review 2026, renewables overtook coal last year for the first time in 100 years, generating 33.8 percent of global power generation  compared with coal’s 33 percent. The review also found that solar power alone accounted for 75 percent of global electricity demand growth that year.

At the same time, the infrastructure powering artificial intelligence is emerging as one of the fastest-growing sources of electricity demand in the global economy. The International Energy Agency’s base-case scenario projects global data center electricity demand reaching roughly 1,200 terawatt-hours annually by 2035, while high-growth projections approach 1,700 terawatt-hours. That would place AI-related electricity consumption near or above the total annual power consumption of countries such as India or Russia today.

Researchers warn that many electrical grids are not expanding renewable capacity and power networks quickly enough to absorb the surge. The result is a mounting global dilemma. The same technology sector promising breakthroughs in medicine, science, and productivity may also deepen dependence on fossil fuels at a critical moment for climate policy.

AI’s growing power demand

The scale of AI-related electricity demand has shifted dramatically within just a few years. Earlier forecasts assumed gradual growth in data center energy use tied to cloud computing and digital services, but generative AI has accelerated those expectations at remarkable speed.

Large AI models rely on massive computing clusters built around specialized chips that consume substantial amounts of electricity and generate significant heat, requiring extensive cooling systems that further increase energy demand. As governments and technology companies race to expand AI infrastructure, competition is extending beyond semiconductors and software talent to include access to reliable electricity generation itself.

The effects are already becoming visible across multiple regions. Gulf states, Southeast Asian economies, the United States, and several European countries are aggressively investing in data center expansion as they attempt to position themselves within the next phase of the global digital economy.

In Ireland, data centers accounted for roughly 21 percent of national electricity consumption in 2023, according to the country’s Central Statistics Office. Singapore previously imposed temporary restrictions on new data center construction over concerns surrounding electricity and water use. In Northern Virginia, one of the world’s largest data center hubs, utilities have warned that surging AI-related demand is complicating long-term electricity planning.

Big Tech versus climate goals

Technology companies have spent years presenting themselves as leaders in corporate sustainability through renewable procurement agreements and net-zero emissions targets. AI is beginning to test the credibility of many of those commitments.

According to BloombergNEF, Bloomberg’s research and analysis division four companies, Amazon Web Services, Google, Meta, and Microsoft, control roughly 42 percent of U.S. data center capacity, giving them enormous influence over regional infrastructure and electricity markets. AWS alone plans to expand from approximately 3 gigawatts of capacity today to nearly 12 gigawatts in coming years.

The challenge is not simply total electricity demand, but the pace at which it is emerging. Renewable energy projects and grid upgrades often require years of permitting and construction. AI infrastructure investment cycles operate much faster because companies are racing to establish dominance in an intensely competitive market.

That mismatch is forcing many utilities back toward natural gas generation because it can be deployed more rapidly than large-scale renewable and transmission projects. Researchers caution that while renewable electricity continues expanding globally, the speed of AI growth risks outpacing the clean energy transition in several major economies.

AI’s electricity race

Competition over artificial intelligence is increasingly becoming competition over electricity availability and infrastructure capacity.

Governments now recognize that AI leadership may depend as much on power generation and grid reliability as software development itself. An April 2026 Brookings Institution analysis warned that access to large-scale electricity infrastructure is rapidly becoming a strategic requirement for AI development, particularly as countries compete to attract energy-intensive data center investment.

Countries with relatively cheap and abundant electricity are becoming attractive destinations for future AI infrastructure investment. The Gulf region illustrates this trend particularly clearly. Saudi Arabia and the United Arab Emirates are simultaneously investing in renewable energy, natural gas infrastructure, and AI development as they attempt to position themselves as global technology hubs. Similar strategies are emerging across parts of Asia where governments view AI as central to long-term economic competitiveness.

Yet the infrastructure burden is creating political tensions as well. Communities hosting large data center projects have raised concerns over rising electricity prices, land use, water consumption, and grid reliability. Utilities are increasingly confronting difficult questions about who should bear the cost of massive infrastructure upgrades designed largely to support private technology firms. In the AI era, access to electricity is becoming a strategic advantage.

The limits of efficiency

Technology firms frequently argue that future AI systems will become substantially more energy efficient. Researchers broadly agree that computing efficiency is improving, but many caution that those gains may not reduce total electricity consumption overall.

Historically, improvements in technological efficiency often lowered costs and expanded adoption rather than decreasing resource use. As AI becomes cheaper and more accessible, businesses may integrate it into a far wider range of economic activity, from logistics and finance to healthcare, manufacturing, and consumer services.

That possibility is forcing policymakers into a difficult position. Governments view AI as economically and strategically essential, yet the infrastructure supporting it may complicate climate goals at precisely the moment global emissions reductions are becoming more urgent. Whether AI ultimately accelerates or undermines climate progress may depend less on the technology itself than on how quickly countries can expand low-carbon energy systems capable of sustaining its enormous power demands.