The rise of artificial intelligence is not just transforming technology—it’s reshaping the global energy landscape. At the heart of this shift are data centres, the silent giants powering everything from AI training models to real-time assistants. According to the International Energy Agency (IEA), global electricity demand from data centres is expected to more than double by 2030, reaching 945 terawatt-hours annually—up from 415 TWh in 2024. The primary driver behind this surge is the explosive growth of AI applications, which rely on unprecedented levels of computational power.
The United States is forecast to bear the brunt of this increase, with data centres expected to account for nearly half of all electricity demand growth by the end of the decade. Japan may see an even sharper impact, with data centres responsible for over 50 percent of that growth. In Malaysia, the figure could be as high as one-fifth—an outsized share relative to its economy. These numbers highlight a looming challenge for energy security and climate goals. As grids are pushed to their limits, countries must figure out how to balance digital progress with environmental responsibility.
What’s especially striking is the role of AI in this transformation. The IEA estimates that AI-specific data centres could quadruple their electricity consumption by 2030. In advanced economies like the US, AI may soon require more energy than the entire heavy industry sector—surpassing even steel, cement, and chemical production combined. This shift marks a profound turning point: power consumption is migrating from traditional manufacturing to the digital core of society. Where chimneys once billowed smoke, server racks now hum—and they’re demanding more power than ever before.
Yet the future doesn’t have to be unsustainable. Experts are calling for a layered strategy to curb this growing energy appetite. First, the development of more efficient AI models—capable of delivering the same performance with fewer computational resources—could dramatically reduce power use. Second, building greener, more energy-efficient data centres is key. And third, switching to renewable energy sources, such as solar and wind, will be essential to making AI infrastructure more sustainable. The IEA also stresses the importance of smart grids that can flexibly adapt to spikes in demand and distribute power more efficiently.
In parallel, researchers are racing to innovate in cooling technologies. A large share of data centre energy isn’t used for computation at all—it’s spent on cooling. Improving how we keep servers at the right temperature could yield enormous energy and water savings, making a significant dent in the environmental footprint of digital infrastructure. Sustainability, in this context, starts not with green marketing slogans but with smarter hardware engineering.
What’s becoming clear is that AI is emerging as one of the dominant forces in modern energy consumption. Where software was once thought of as ethereal and immaterial, the AI era is reminding us that intelligence requires infrastructure—and infrastructure needs power. The next few years will be defined not just by breakthroughs in machine learning, but by innovation in energy efficiency. In the race to shape the future, those who master both will lead.
