How much energy should society be willing to spend on Artificial Intelligence?

Artificial intelligence is becoming part of everyday life more and more. Students use it to understand difficult homework questions, businesses rely on it to complete tasks faster, and millions of people turn to chatbots for information for any questions they may have. However, behind every AI response is a network of powerful computers that requires electricity to operate. As the technology expands, an important question emerges: how much energy should society be willing to spend on artificial intelligence?

The growing demand for power

According to the International Energy Agency (IEA), electricity consumption from data centres increased by 17 per cent globally in 2025. Data centres focused on AI saw even faster growth, with electricity consumption rising by approximately 50 per cent. These facilities have the computer systems responsible for processing information, running online services, and powering AI. As more people and businesses adopt these tools, the infrastructure supporting them must also expand further increasing the consumption of electricity.

In its 2026 analysis, Key Questions on Energy and AI, the IEA projects that total data-centre electricity consumption could nearly double, from 485 terawatt-hours in 2025 to 950 terawatt-hours by 2030. This represents a challenge for energy systems already expected to meet increasing demand from homes, transportation and industry, and now from AI.

Convenience comes with heavy costs

For students, asking an AI chatbot to explain a mathematical equation may seem like a simple digital interaction. Yet AI services depend on physical infrastructure, including servers, cooling systems and electricity networks. The environmental impact goes beyond the screen, even though the energy required for individual tasks varies considerably.

The IEA also highlights an important difference: energy efficiency is improving, and a simple text-based AI query most of the time consumes less electricity than running a television over the same period. However, more demanding applications, including AI-generated videos and complex reasoning tasks, can require substantially more energy. Therefore, the problem is not simply that every AI question consumes enormous amounts of electricity. Rather, it is the combination of increasing usage, more powerful applications and expanding infrastructure that deserves attention.

Who should take responsibility?

Technology companies have a responsibility to consider the environmental consequences of their products. Improving hardware efficiency, using renewable electricity and designing more efficient AI systems could help limit the impact of growing demand. Governments and energy providers must also consider how new data centres will affect electricity supply, affordability and long-term infrastructure planning.

According to UNESCO, artificial intelligence offers opportunities to improve education and expand access to knowledge, but its development must also account for broader social responsibilities. This principle applies to environmental sustainability as well: technological progress should benefit people without overlooking its wider consequences.

A smarter way forward

For young people growing up with AI, the answer should not necessarily be to stop using the technology. Artificial intelligence can support learning, scientific research and problem-solving when used thoughtfully. Instead, society should encourage innovation while demanding greater transparency about its environmental footprint.

The future of AI will depend on more than how quickly computers can generate answers. It will also depend on whether the companies building these systems can make them more efficient, sustainable and accessible.