Artificial intelligence is transforming almost every part of the economy.
From automated customer service and predictive analytics to scientific research and generative AI, the technology promises enormous opportunities for businesses and society.
But behind every AI query, model and digital service sits something very physical: data centres.
And in August, new analysis has highlighted just how significant their environmental footprint could become in the UK.
Two planned data centres in England could collectively generate more than 4.5 million tonnes of carbon emissions every year, according to analysis by environmental organisation Foxglove. That figure would exceed the UK emissions attributed to ExxonMobil in 2023.
The projects are expected to require around 1.3GW of electricity.
That creates a fascinating sustainability dilemma.
We cannot simply electrify our way out of the problem
Data centres are central to the digital economy, and demand for computing capacity is growing rapidly as AI adoption accelerates.
But electricity demand matters.
The UK is working to decarbonise its electricity system, while simultaneously trying to accommodate new and increasingly energy-intensive industries.
In the case of the two proposed data centres, concerns have been raised because lengthy grid connection times could lead to the facilities using on-site gas-fired power generation.
That creates a paradox.
AI is often presented as a technology that can help businesses optimise energy use, reduce waste and improve sustainability.
But the infrastructure required to run AI can itself consume huge amounts of energy.
The sustainability question businesses should be asking
This isn’t an argument against AI.
The technology has enormous potential to support sustainability.
AI can help organisations:
- Identify energy inefficiencies.
- Optimise logistics and transport.
- Forecast demand.
- Reduce material waste.
- Monitor environmental data.
- Improve renewable energy management.
- Analyse complex sustainability datasets.
But businesses need to consider the whole lifecycle and infrastructure footprint of digital transformation.
If an organisation announces that it is becoming more sustainable because it has digitised processes or introduced AI, the conversation should not end there.
Where does the computing happen?
How much energy does it consume?
What is the source of that electricity?
How efficiently is the data centre operating?
What happens to the hardware at the end of its useful life?
And how much additional digital infrastructure will be required as usage grows?
Green AI needs more than green claims
There is a growing opportunity for businesses to think about responsible AI and sustainable digitalisation together.
The UK’s transition to a low-carbon economy will require more electricity, more digital infrastructure and more computing power.
That makes energy efficiency and renewable generation increasingly important.
It also means sustainability teams will need to work more closely with IT, procurement and finance departments.
The sustainability conversation is no longer confined to manufacturing plants, vehicles and buildings.
It is increasingly happening inside servers, networks and cloud infrastructure.
AI may help us solve some of the world’s biggest sustainability challenges.
But if we want that future to be genuinely sustainable, we need to make sure the technology itself is being built on sustainable foundations.
The question isn’t whether AI can help us become more sustainable.
It’s whether the growth of AI can itself become sustainable.