The global conversation on data centre growth is getting louder — and for good reason.
Most analysts expect the number and size of data centres worldwide to continue climbing sharply this decade. With that growth comes a surge in electricity demand that is already reshaping energy forecasts.
The International Energy Agency (IEA) warns that in the United States alone, power consumption by data centres is set to account for almost half of all growth in electricity demand between now and 2030. Much of this is being driven by AI workloads. By the end of the decade, the IEA projects that processing data will consume more electricity in the US than manufacturing all energy-intensive goods combined — aluminium, steel, cement, and chemicals. Globally, the IEA is predicting that data centre energy consumption will grow to 945TWh per year by 2030, from 415TWh in 2024.
Goldman Sachs Research paints a similar picture: global data centre power demand could rise by 50% by 2027 and as much as 165% by 2030 compared to 2023. In a recent report, the company estimates that the current power usage by the global data centre market to be around 55 gigawatts (GW). This is comprised of cloud computing workloads (54%), traditional workloads for typical business functions such as email or storage (32%), and AI (14%). Its future demand modelling estimates for these workload types suggest power demand will reach 84 GW by 2027, with AI growing to 27% of the overall market, cloud dropping to 50%, and traditional workloads falling to 23%.
Other analysts and investment banks are equally bullish (or worried, depending on your perspective) about the scale of the challenge.
The UK Government’s new report: a calmer picture
Against this backdrop, the UK Government’s Department for Energy Security and Net Zero has just published a commissioned report by Europe Economics on The Impact of Growth of Data Centres on Energy Consumption.
The study uses a novel methodology to compare the electricity used for certain digital activities with that used for their physical equivalents – for example:
• Reading an e-book vs reading a printed book
• Streaming a film vs watching it on a Blu-ray disc
• AI translation vs human translation in an office
In all three cases, the results suggest that the digital option uses the same or less electricity than the physical one and in some cases dramatically less. An e-book download, for instance, is estimated to use hundreds of times less UK electricity than printing a physical book.
From these examples, the implicit message is that digitalisation can be more energy-efficient than traditional physical activities, perhaps even a contributor to net zero goals.
So… who’s right?
Here’s the problem: the UK report isn’t actually contradicting the IEA or Goldman Sachs, it’s just answering a different question.
The Government’s study measures direct electricity use for one “like-for-like” activity in a hypothetical world where digital doesn’t increase total demand. The IEA and Goldman Sachs are looking at total real-world demand growth, which includes the fact that digitalisation doesn’t just replace physical activities — it creates new ones and massively scales up how often we do them.
Put simply, the UK report is like saying, “A single LED bulb uses less power than a candle.” but the global forecasts are saying, “Yes, but now everyone’s leaving the lights on 24/7 in ten more rooms.”
Why the difference matters
The UK study’s narrow scope means it ignores the rebound effect, where people consume far more when services are digital and instant. It only counts UK electricity use, meaning energy consumed abroad in manufacturing devices or powering overseas data centres is excluded, and it leaves out the embodied energy required to produce and maintain the devices, networks, and data centres themselves.
The analysis also focuses on a handful of relatively low-energy digital activities rather than the high-intensity AI and cloud workloads expected to drive most future growth. While this makes the study useful for very specific policy questions, it becomes risky if policymakers or media headlines take it as evidence that data centres aren’t a significant energy issue. These cases don’t capture high-energy digital activities with no real physical equivalent (e.g. AI model training, crypto currency, cloud gaming, generative video).
The real takeaway
Both perspectives are technically correct within their own frame of reference – many digital services are more energy-efficient per task than their physical equivalents. But in the real economy, the scale and speed of digital adoption – especially with AI – are pushing global electricity demand from data centres to unprecedented levels.
Ignoring this truth risks sleepwalking into an energy crunch where supply can’t keep up with the digital services we’ve come to depend on.
At vXtream, as a data centre infrastructure provider, we are acutely aware of our responsibility to minimise energy consumption and operate as efficiently as possible. We implement industry best practices where we can, such as cold aisle containment to optimise cooling performance, reduce wastage, and maintain consistent environmental conditions.
We also prioritise sustainability in our operations by procuring 100% green energy, ensuring that the power driving our infrastructure is sourced from renewable supplies. Our commitment is to deliver high-performance digital services while actively reducing our environmental footprint.
And we don’t greenwash. Please don’t hesitate to contact us to find out more.
Image by Aliis Sinisalu on Unsplash
Comments are closed.