Hassabis: AI Will Fix Climate, Silent on Data Centre Power

Why does Hassabis say AI can solve climate change?

Hassabis argues AI can deliver breakthroughs in climate and weather modelling, plasma control in fusion reactors, superconductors, better batteries and solar materials. In his view, AI itself will produce the technology that solves the climate problem.

Demis Hassabis, Google DeepMind's co-founder and board chairman, spoke at the International AI Summit in Dublin. Addressing ministers, press and delegates, he walked through AI's record in science and mathematics before turning to climate.

Hassabis conceded that AI is progressing more slowly in physics, chemistry, biology and engineering than in maths or the game of Go. He attributed this to those disciplines needing 'real-world loops' where forming a hypothesis depends on human creativity, intuition and curiosity.

In Hassabis's framing, the hardest part of science is asking the right question. He called that step 'even harder than finding the answer' and implied AI cannot yet replace humans there.

• Climate and weather modelling: flood, wildfire and hurricane forecasting
• Plasma control in fusion reactors
• Superconductors and new materials
• More efficient batteries and solar panels

Why did Hassabis never mention AGI in Dublin?

Hassabis did not say the words artificial general intelligence once in his 20-minute Dublin talk. Five months earlier at Google I/O he described AGI as 'the foothills of the singularity' and said it was on the horizon, making the silence notable.

In May 2026 at Google I/O in California, Hassabis spoke at length about AGI, the concept of AI equivalent to human intelligence. Five months later, offering no progress report on AGI suggests the industry's focus has shifted to AI's scientific promises.

Instead, Hassabis spent his time on AI's proven record in mathematics, coding and games. The choice signals DeepMind's recent tilt toward scientific applications.

In 2024 Hassabis and DeepMind's John Jumper shared the Nobel Prize in Chemistry with David Baker of the University of Washington for an AI model that predicts protein structures. That award remains the clearest evidence of the company's science-first strategy.

Why is AI's environmental impact under scrutiny?

Demand from AI data centres is accelerating a shift toward fossil-fuel power plants, which are a primary driver of the climate crisis. That is why AI's environmental balance sheet is being questioned more and more.

In Dublin, Hassabis was asked about AI's serious sustainability problem. Enormous, power-hungry data centres increase strain on grids and consumption of natural resources.

Rather than acknowledge that pressure on the grid and natural resources, Hassabis emphasised AI's potential in green technology, clean energy and disaster forecasting. The choice mirrors the industry's standard response to environmental criticism.

Data centre investment is climbing fast in Turkey too. Unless the electricity demand of these energy-intensive facilities is matched by growth in renewable capacity, AI's carbon balance will stay contested. A campaign against a data centre the size of 57 football pitches in Wales is a concrete example of that backlash.

• Data centres increase demand for fossil fuels
• Hassabis did not answer that criticism directly
• Companies argue for breakthroughs in clean energy and materials
• The race: energy demand versus green technology

What does Google's new Gemini 4 Argon model bring?

Hassabis briefly mentioned Gemini 4 Argon, the model Google released the same week as the summit. Details were limited; he focused instead on scientific applications and his climate pitch.

Google announced Gemini 4 Argon the same week as the summit. Hassabis gave the release a single sentence and no technical detail.

Around the same period, debate was also swirling over a claim that OpenAI had solved the Navier-Stokes equations. Hassabis touched on that only superficially.

Making the climate promises concrete would require independent evidence of how far the model is used in energy and materials research. No comprehensive published data exists on that yet. Google, meanwhile, keeps pushing AI into hardware, as with Siri AI arriving on older Apple TV and HomePod devices.

How much energy do AI data centres use, and how can the problem be fixed?

AI data centres burn heavy amounts of electricity for cooling and computation, and in some regions demand is high enough that fossil-fuel plants are being brought back online. Fixing it requires renewable power, efficient cooling and model optimisation.

Data centre energy demand is rising in step with the growth of AI models. Most of these facilities still run on fossil fuels, which enlarges AI's carbon footprint.

Some companies in the sector are turning to clean sources such as nuclear and geothermal. But the scale of those investments does not yet match the growth in demand.

Data centre investment has also accelerated in Turkey. Meeting these facilities' energy needs with renewable capacity is critical to the country's climate targets.

1. Power data centres with renewable energy.
2. Move to efficient cooling such as liquid cooling.
3. Cut compute needs by optimising model training.
4. Install energy recovery systems.

Bottom line

Hassabis's Dublin message was clear: AI will be a key technology in solving the climate crisis. Yet data centre energy demand is growing far faster today than the promised benefits. Anyone watching AI and data centre investment in Turkey should track where these facilities get their power and what renewable capacity is planned. Independent research over the next two years will show how real the climate promise is.