Artificial intelligence might be the key to solving one of its biggest problems: energy consumption. As AI systems demand more electricity, a computer scientist at the University of Auckland argues that the same technology could accelerate innovations in renewable energy and chip efficiency, creating a self-correcting cycle that mitigates AI's environmental footprint.

A new report from the United Nations University warns that AI could consume three percent of the world's electricity by 2030, a staggering increase from current levels. Rather than viewing this as a dead end, Dr. Ulrich Speidel, a computer science lecturer and technology consultant, sees an opportunity for AI to drive the very advances needed to power itself more sustainably.

How Can AI Help Reduce Its Own Energy Footprint?

Speidel identified three specific areas where artificial intelligence could accelerate efficiency gains:

• Solar Panel Efficiency: Current solar photovoltaic cells convert only about 25 percent of the energy from sunlight into electricity. AI could help researchers optimize panel designs and materials, allowing existing solar farms to generate significantly more power without expanding land use.
• Battery Technology Development: Rather than building and testing countless battery prototypes, AI can rapidly simulate different chemical combinations and configurations. This computational approach could dramatically speed up the discovery of next-generation battery technologies with higher energy density and longer lifespans.
• Semiconductor Optimization: AI can process vast numbers of potential designs to make computer chips more energy-efficient. Chips developed a decade from now could consume a fraction of the power required by today's processors, directly reducing the energy demands of AI systems themselves.

"If you've got access to a tool like AI that's actually able to crunch a large amount of numbers and able to go through a lot of potential technologies and the potential ways of making semiconductors more efficient, then again, that should accelerate hopefully the development of semiconductor technology," said Dr. Ulrich Speidel.

Dr. Ulrich Speidel, Computer Science Lecturer and Technology Consultant, University of Auckland

Why Does This Matter for AI's Future?

The energy consumption of AI is not merely an environmental concern; it's becoming a practical bottleneck for the technology's expansion. Data centers powering large language models and other AI systems consume enormous amounts of electricity, raising questions about sustainability and operational costs. If AI can help solve the efficiency problems in energy generation and storage, it creates a positive feedback loop where the technology becomes progressively more sustainable.

Speidel's perspective offers a counterpoint to purely pessimistic assessments of AI's environmental impact. Rather than accepting that AI will inevitably drain resources, he suggests the technology's computational power could be redirected toward solving the very problems it creates. This approach doesn't eliminate the challenge but reframes it as a solvable engineering problem where AI itself becomes part of the solution.

The timeline matters here. With AI electricity consumption projected to reach three percent of global usage within four years, efficiency improvements in solar cells, batteries, and semiconductors would need to scale quickly to make a meaningful difference. Speidel's argument is that AI's ability to rapidly model and test new designs could compress what might otherwise take decades of traditional research into a much shorter timeframe.

Whether this optimistic scenario materializes depends on whether the computational resources currently devoted to training larger and larger AI models can be redirected toward efficiency research. It also requires coordination between AI developers, energy companies, and materials scientists to prioritize these efficiency gains alongside other AI development goals.