Hassabis Says AI Will Deliver the Technology to Tackle Climate Change

Google DeepMind co-founder Demis Hassabis told an audience of global ministers in Dublin on Wednesday that AI itself may be the key to solving the climate crisis — even as the industry's data center boom drives new demand for fossil-fuel power. "I actually think with the help of AI, that's what's going to end up giving us the technology to tackle the climate question," he said.

The remark came at the International AI Summit in Ireland, where Hassabis spoke in a 20-minute slot on the main stage, following Executive Vice President of the European Commission Henna Virkkunen. His audience included government ministers, media and delegates. The stakes are concrete: the massive push to build sprawling, energy-hungry data centers is spurring a surge in demand for fossil-fuel power plants, which remain a key driver of the climate crisis.

What did Hassabis actually say?

His talk covered innovation in science and math, frontier AI and — on a hopeful tack — AI's potential to address climate issues. He briefly mentioned this week's rollout of Google's new Gemini 4 Argon model and the debate around OpenAI's claim to have solved the Navier-Stokes equation, one of history's most difficult mathematical problems.

But the most telling detail was what he left out. AGI was not mentioned once.

That's a sharp turn from May, when Hassabis spoke at length at Google I/O in Mountain View, California, about artificial general intelligence — the hard-to-pin-down concept of AI matching human intellectual capabilities. He called current progress the "foothills of the singularity" and estimated that achieving AGI is on the horizon. Five months later, a status report on that goal seemed to have slipped his mind.

Where does AI still fall short, according to DeepMind?

Instead of forecasting AGI, Hassabis outlined where AI excels today: mathematics, coding, and games like Go and the classic Atari titles. Then he drew a contrast that framed the rest of his argument. Physics, chemistry, biology and engineering, he said, require "real-world loops" — human creativity and judgment, intuition and curiosity in forming hypotheses and developing lines of inquiry.

He has the credentials to make that case. In 2024, Hassabis and DeepMind colleague John Jumper, together with University of Washington scientist David Baker, won the Nobel Prize in Chemistry for their work on an AI model that can predict protein structures.

His view of where machines struggle is surprisingly philosophical. "I think that's the hardest thing in science is actually asking and articulating the right question," Hassabis said. "That's even harder than finding the answer."

How did he address AI's own energy problem?

Before exiting the stage in Dublin, Hassabis was asked directly about AI's significant sustainability footprint. He did not acknowledge the stresses data centers place on energy grids and natural resources. Instead, he expressed optimism about AI driving breakthroughs in green technology, clean energy and models for predicting natural disasters like floods, fires and hurricanes.

"The most important thing is that the AI systems we're building, if we build the right types, should be able to help with things like climate modeling and weather modeling," he said.

He then extended the timeline. "Looking further forward," he said, "AI systems will help with new sources of energy — fusion plasma containment for fusion reactors, new materials like superconductors, better batteries and better solar materials. I actually think that with the help of AI, that's what's going to end up giving us the technology to tackle the climate question."

Why the silence on AGI matters

The missing AGI talk is the story beneath the story. In May, Hassabis framed superhuman AI as imminent enough to discuss at a developer conference. In Dublin, before regulators and ministers, the framing shifted to scientific tools and climate solutions — domains where DeepMind already has a Nobel-winning track record.

For policymakers weighing data center permits and energy policy, the pitch was clear: tolerate the industry's power consumption now, and AI will pay the climate debt back later through fusion, better batteries and disaster prediction. Whether that bet pays off is the question Hassabis left unanswered on stage.