“I don’t think we can do both.” That’s what Michael Simmonds, Economics Editor at the Spectator, said in the publication’s podcast. “We can push ahead with the infrastructure we need if we’re to utilise AI to boost growth,” he continues, “or we can stay on the path to net zero.” 

Maintain the UK’s status as an AI pioneer? Or give that up and aim for net zero instead? This is being framed as a binary choice by not just Simmonds, but many others. The Centre for British Progress wrote in its ‘Unblocking AI Growth Zones’ report that we should opt for ‘gas now, grid and nuclear later’ in order to advance AI. To remain non-partisan, here’s the perspective of Reform’s Deputy Leader Richard Tice, who said that, “Net Stupid Zero means we cannot benefit from AI data centers like US & nations with low energy costs.”

However, the idea that AI and net zero can’t coexist is a misconception. When we build AI-ready data centres that reduce emissions instead of adding to them, our AI use can become a stepping stone that helps us reach net zero.

How legacy data centres are damaging our move towards net zero

It’s easy to see why many people believe that AI growth will hinder our progress towards net zero. Legacy or hyperscale data centre facilities counteract our climate goals by taking over green space, releasing emissions, and draining water supplies.

We spoke to Daniel Tremayne-Pitter, CEO at Dark Matter, a consultancy that educates governments and brands on emerging technology such as AI and data centres. He agreed that, “The idea of AI and net zero seems oxymoronic in every possible way.”

He explained why this misconception exists: “There’s a gap in the narrative,” he says. “Information has filled that void, and is educating the media, public and government to believe that [AI and net zero cannot coexist].”

Where does that information come from? Outdated reports from the data centre industry that don’t fit our present-day context, according to Daniel. They represent information “from a completely different region at a different time,” Daniel says. In other words: legacy, hyperscale data centres. 

Here’s the crucial distinction: not all data centres are built the same, and data centres can flip the script and help, not hinder, our net zero transition.

Introducing a different kind of data centre

So, what sort of data centre could help us transition towards net zero? “Modern design, modern architecture, and innovative thinking about data centre placement,“ Daniel says. “There’s a world where energy and resources can be carefully managed to still deliver all of the AI outcomes that we want, but in a more considerate way.”

Deep Green’s AI-ready data centres are designed with net zero goals, reduced carbon emissions, and community impact in mind. They’re planned around facilitating AI growth with high-density capacity and supporting the environment and the local community.

Firstly, they consider location. By building into existing infrastructure, they don’t take over green space compared to legacy facilities. By choosing a location near heat offtakers (i.e. public or residential buildings that can make use of excess heat), they are able to donate heat from their servers to warm public spaces, businesses and homes.

This heat donation has a triple benefit. The heat offtakers reduce their heating emissions and costs. In turn, this reduces strain on the UK’s grid,  as the heat offtakers reduce their energy requirements. Finally, the data centre reduces its cooling emissions, as cooled water is returned from the heat offtaker that contributes to cooling the data centre servers.

When it comes to cooling, choosing a closed-loop or adiabatic cooling system reduces a data centre's water usage to as little as a domestic house per year. The notion that data centres will drain our water supplies amidst the UK’s increasing heatwaves is only true of legacy facilities with consumptive cooling systems.

Data centres compared: why we can have AI and net zero

The government policy we need for an AI-forward, net zero transition

So, if it’s possible to sustain AI growth while still progressing towards net zero, why isn’t the UK government ensuring we only build the data centres that can make that happen? 

Policy must distinguish between data centres that contribute to the grid and reduce emissions, and those that drain the grid and release emissions. By accelerating planning permission and grid access for data centres that reuse heat, the UK can maintain AI growth while still transitioning towards net zero.

The data centres we build now will impact our climate for decades

With the Climate Change Committee’s (CCC) Seventh Carbon Budget stressing the need to accelerate grid build-out, it’s clear that grid strain is a government priority. Meanwhile, the UK’s current target is to reach net zero by 2050. While that might sound far in the future, the data centres that we build now will have a long-term effect on whether we can achieve this goal.

Data centres stick around for decades. Retrofitting existing data centres is costly and time-consuming, while connecting them to heat networks retroactively is often impossible due to location or planning constraints. We must allocate our scarce power and planning bandwidth to data centres that contribute towards net zero, rather than on a first-come, first-served basis. Otherwise, we’ll be stuck with legacy facilities for the long term.

The blueprint: how AI and net zero work together

So, what does a modular, AI-ready data centre look like? In Urmston, our 400kW facility donates heat to move Urmston Leisure Centre, saving Trafford Leisure around £80,000 a year on community heating costs, reducing CO2 emissions by 100 to 150 tonnes, and relieving pressure on the grid. Its closed-loop cooling system uses practically zero water. What’s more, it earns its social license to operate through community programmes such as teaching locals to swim and funding tree planting in the nearby area. Want to see this blueprint in action? Check out our Urmston facility..