All kinds of companies and countries are racing to build the best AI models, but in the long run, the determinant to success in the AI space could be just one factor — energy.
Demis Hassabis, the CEO of Google DeepMind and a leading figure in artificial intelligence research, recently shared his perspective on one of the most pressing challenges facing the AI industry. In a discussion with CNBC, Hassabis acknowledged the severe constraints the industry faces and painted a picture of a future where energy availability becomes the fundamental currency of artificial intelligence development. His comments come at a time when the AI industry’s enormous energy demands are drawing increasing scrutiny from policymakers, industry leaders, and environmental advocates alike.
Hassabis began by addressing the immediate physical limitations facing AI development. “There’s lots of physical constraints. So of course, no one ever has enough chips. And we are lucky that we have our own TPU range in addition to GPUs, but there just aren’t enough compute chips in the world really for the demand. And of course, in the end, that comes down to energy as well,” he said.
The DeepMind CEO then articulated a striking vision for the relationship between energy and intelligence in the age of advanced AI. “There’s this idea of energy. What we effectively is synonymous with intelligence as we get into the era towards AGI,” Hassabis explained. “Now the interesting thing is I think that AI itself will help here in the sense of getting more efficiencies out of existing infrastructure, but helping with things like material design, better solar materials.”
Hassabis pointed to concrete examples of how AI could help solve its own energy problem. “But it could also help with new breakthrough technologies like fusion. We have a collaboration with Commonwealth Fusion in the US to help contain plasma in fusion reactors. And one of my pet projects is can we come up with a room temperature superconductor material using AI? So I think there are multiple breakthroughs that AI could come up with and help us come up with that would help with the energy situation. In fact, indeed that’s, I think that’s one of the most promising use cases of AI,” he said.
The Google DeepMind CEO also highlighted the progress being made on the efficiency front. “And then the other thing is as these systems are getting better, they’re also getting 10x more efficient per year. So if you look at our range of models, we have our kind of lighthouse model, our pro versions of Gemini, but then we have our Flash versions, which are way more efficient and they’re sort of workhorse models that we use for everything. And they use techniques like distillation where you have a big model that teaches a smaller model. And the smaller models are really, really efficient. And I think there are more and more innovations and techniques like that that will keep bringing the efficiency curve down. And so you get much better performance per watt,” Hassabis concluded.
Hassabis’s comments underscore a growing recognition across the tech industry that energy infrastructure will be the defining constraint—and competitive advantage—in the race to build advanced AI systems. The scale of the challenge is staggering. Former Google CEO Eric Schmidt has warned that AI could end up consuming a massive portion of the world’s electricity, while OpenAI alone plans to use as much energy as half of India’s installed capacity within eight years.
This energy bottleneck is already reshaping geopolitical dynamics in AI development. Access to abundant electricity has emerged as a potential advantage for China in its competition with the United States, as the country rapidly expands its power generation infrastructure. Meanwhile, Microsoft CEO Satya Nadella has argued that AI’s high energy consumption means its outputs will need to be socially useful to justify the resource expenditure.
What makes Hassabis’s perspective particularly noteworthy is his dual emphasis on both the problem and potential solutions. By positioning AI as both the driver of unprecedented energy demand and the potential key to breakthrough energy technologies—from fusion reactors to room-temperature superconductors—he suggests a path forward that doesn’t require choosing between AI advancement and energy sustainability. Whether this optimistic vision materializes, however, may well determine not just which companies and countries lead in AI, but whether the technology can scale sustainably at all.