Can data centres really be green?

The UK’s race to become a global data centre superpower is colliding with another national priority – reaching net zero. The government has made it clear that it sees digital infrastructure as essential to economic growth, but the scale of electricity demand now projected from the sector has forced a difficult question into the spotlight: Can data centres ever truly be green?

 

The debate has sharpened in recent weeks, with reports suggesting more than 100 UK data centres are planning to use gas generation to support rising power needs. And this isn’t just a regional issue; it’s playing out worldwide  – in the US, Elon Musk’s xAI has reportedly expanded gas-powered infrastructure without permits.

 

At first glance, it looks like a direct contradiction of the tech sector’s sustainability promises. An understandable concern given data centre electricity demand in the UK could rise from around 7.6TWh today to somewhere between 30 and 70TWh by 2050. That’s no marginal increase. It’s a step change the system wasn’t designed around, and it’ll place real pressure on existing net-zero pathways.

 

 

The looming energy challenge

If data centres continue operating as conventional and inflexible industrial loads, the consequences for the energy system are obvious. Greater peak demand will mean more pressure on the grid, increased reliance on gas generation during periods of low renewable output, and higher overall system costs.

 

Under that model, the expansion of digital infrastructure risks undermining decarbonisation efforts, rather than supporting them.

 

But framing data centres purely as an energy problem misses something important. Unlike many industrial users, they have a degree of operational flexibility that’s still largely poorly understood. The real question is not whether data centres consume large amounts of electricity - they clearly do - but whether the sector can evolve from being a passive consumer of power into an active participant in the energy system itself.

 

And that distinction matters enormously.

 

 

AI is changing the energy equation

The common perception is that data centres require rigid, constant demand 24 hours a day. In reality, their load profiles are often far more dynamic. Although idle load can account for significant capacity, certain workloads - particularly AI model training, data backups and batch processing - can be shifted in time or even location, and total demand can fluctuate substantially over relatively short periods.

 

This creates an opportunity that’s barely begun to be exploited. If operators can respond intelligently to grid conditions, delaying non-urgent computing tasks when clean generation is low and accelerating them when it’s abundant, they become part of the balancing mechanism that the future electricity system increasingly depends upon.

 

Google has already experimented with dynamically moving workloads based on grid carbon intensity. Combined with battery storage, sophisticated control systems and more advanced power purchasing agreements, this type of flexibility could materially reduce the system’s reliance on carbon-intensive generation during periods of stress.

 

That’s where the concept of a genuinely green data centre begins to move beyond marketing language.

 

 

Beyond greenwashing and carbon accounting

At present, much of the sustainability conversation remains dominated by accounting mechanisms rather than operational reality. Most corporate “green power” claims still rely heavily on renewable energy certificates and offsetting frameworks. In the UK, Renewable Energy Guarantees of Origin (REGOs) are frequently used to badge electricity as renewable, even though the physical power consumed still comes from the same mixed national grid.

 

Some suppliers have attempted to improve this through half-hourly matching frameworks, aligning renewable certificates more closely with actual consumption patterns. While more sophisticated, these approaches do not fundamentally change the underlying reality: fossil fuel generation remains part of the grid mix, particularly during periods of peak demand or low renewable output.

 

The danger is that sustainability becomes a paper exercise rather than a physical outcome.

This is why the distinction between procurement and behaviour is becoming increasingly important. Simply purchasing certificates does little to reduce real-time system emissions if underlying consumption patterns remain unchanged. By contrast, a data centre capable of intelligently shifting workloads, integrating battery storage and responding to live grid signals can actively lower carbon intensity at critical moments.

 

What’s clear is that more honest conversations are needed about what “green” actually means.

 

 

The old industrial energy model is disappearing

The challenge is compounded by broader structural changes within the energy market itself. Historically, large industrial consumers have relied on relatively stable baseload generation and liquid forward energy markets to hedge long-term exposure, but that model is gradually disappearing.

 

As more electricity generation is channelled through mechanisms such as Contracts for Difference (CfDs), and as the system becomes increasingly dominated by intermittent renewables, forward market liquidity is on the decline. The result is a more volatile and less predictable pricing environment for major consumers like data centres.

 

This is one of the defining energy challenges of the AI era. Businesses are simultaneously facing rising electricity demand, increasing price volatility and greater uncertainty around long-term supply security. Traditional hedging strategies are becoming harder to construct just as demand growth accelerates.

 

What replaces the old model is likely to be far more integrated. Future energy systems will increasingly combine grid connectivity, co-located renewables, battery energy storage systems (BESS), flexible demand and intelligent software controls into a coordinated whole.

 

 

From energy consumers to energy assets

Data centres are uniquely positioned within that future architecture. Their ability to rapidly adjust demand, operate across multiple sites and integrate advanced control systems gives them capabilities few other industrial sectors possess. In effect, they can evolve into energy assets in their own right.

 

But none of this happens automatically. Policy will determine whether the opportunity is realised or wasted.

 

At present, there is a growing contradiction at the heart of UK energy policy. On one side, the government is aggressively encouraging data centre expansion and AI infrastructure investment. On the other side, current market structures are making it harder for large consumers to secure predictable long-term energy strategies.

 

Meanwhile, flexibility - arguably one of the most valuable assets in a renewable-heavy electricity system - remains underutilised and inconsistently rewarded, and that needs to change.

 

 

The future of green data centres

The question needs to move beyond whether data centres create challenges for the energy system – they undoubtedly do. The more important question is how they are integrated into it.

 

If they continue to operate as passive demand, they will increase system stress and deepen reliance on gas generation. But if operators, policymakers and energy markets evolve together, data centres could become one of the key enabling technologies of the clean energy transition itself. Now that’s a thought.

 

---

 

Chris Bowden is the founder and CEO of SQE (formerly Squeaky Energy) - an energy supply platform exclusively for industrial and commercial (I&C) users

 

Main image courtesy of iStockPhoto.com and Dragon Claws