Recently, I was in conversation with some fellows from Silicon Valley and Shenzhen — two of the world’s tech powerhouses — swapping stories about our latest AI projects. One of them pulled up a live feed from their lab: A warehouse-sized hall lined with racks of AI servers, each glowing and humming like a beehive in overdrive. In the corner, an engineer pointed to a monitor showing the facility’s power draw — more than an entire city block. For a moment, the conversation went quiet. We all knew what that number meant.In a world already straining its energy grids, AI’s appetite is not just impressive — it is a warning. That night, I could not shake a question: If this is just the beginning, how will we keep the lights on for the AI future? We talk about artificial intelligence as if it is a purely digital revolution — a clean leap from matter into mind. But AI is a physical machine, and physical machines run on energy.

A single large AI model can consume as much electricity as hundreds of homes do in a year. Multiply that across the dozens of tech companies in an arms race to build ever-bigger models, and we are creating a new kind of energy hunger.

This is not just a climate issue. It is a geopolitical one. Energy efficiency in AI will decide who controls the technology, how widely it is shared, and whether we can use it without accelerating global warming. And right now, we are on track to lose that control. We have a narrow window to change course — and it begins........

© The Pioneer