Understanding the world of energy is complex – sometimes it feels paradoxical.
In 2024, for the first time in 50 years, oil accounted for less than 30% of global energy consumption last year, while low-carbon energy now generates over 40% of the world’s electricity. And yet, neither oil nor coal consumption is declining. Greenhouse gas emissions even rose by 0.8% last year.
The Green Consumption Paradox
This has led some to question whether we’re facing a “green consumption paradox”: the idea that the more clean energy we produce, the more overall electricity demand grows to absorb it. In this model, fossil fuel-based plants, particularly coal and gas, remain in use as a fallback, their role never fully eliminated.
A report published today by the International Energy Agency (IEA) explores the impact of AI on our energy systems. At first glance, it seems to reinforce the green consumption paradox: global electricity demand from AI and data centres is expected to rise sharply over the next 5 to 10 years.
Depending on assumptions around energy efficiency in the sector and the speed of its development, demand could more than double by 2030, reaching 945 TWh, the equivalent of Germany’s current electricity consumption. By 2035, this could triple to 1,200 TWh, roughly matching India’s. This kind of growth represents a serious challenge, particularly for countries like the United States, which aim to host a disproportionate share of this emerging industry.
Electricity demand will be shaped by new uses. AI Is just one piece of the puzzle
That said, the picture isn’t entirely bleak.
Global electricity demand in 2023 stood at around 30,000 TWh. If AI drives an increase of 800 TWh by 2035, that’s a rise of just 2.7%. Significant, but manageable.
Global electricity demand from AI and data centres is expected to rise sharply over the next 5 to 10 years.
In fact, other factors are likely to push electricity demand even higher in the years ahead: the mass adoption of electric vehicles, the rollout of electric heating and cooling systems, and the increasing electrification of industry. These trends are transforming how we use electricity and signalling the gradual retirement of fossil fuels in key sectors.
Assuming the current growth in renewable energy continues, this shift could finally begin to push emissions down on a large scale.
The real challenge: matching demand to supply
The real issue isn’t just higher electricity consumption – it’s how and when we use energy. The energy transition means integrating hundreds of millions of connected devices into our power systems. These devices don’t just consume more electricity; they also change the shape of demand over time.
Unlike fossil fuels, renewables don’t produce power on demand. Their output varies by hour, day, and season. Sometimes there’s too much wind and solar, sometimes not enough. The core challenge, then, is not production. It’s flexibility.
To make the most of renewables, we need to shift electricity demand to the moments when clean energy is most abundant.
But expecting consumers to track weather patterns and time their usage accordingly is neither realistic nor scalable, even if moving off-peak hours into sunny daytime slots is a step in the right direction.
So, what’s the solution? Enter ‘Virtual Power Plants’
Part of the answer lies in smart, automated systems, powered by AI. Thanks to new digital tools, it’s now possible to control millions of devices in real time: cheaper smart sensors for real-time measurement, advances in IoT enabling automated control, and powerful AI algorithms hosted in the cloud which flexibly manage electricity demand without disrupting user.
To make the most of renewables, we need to shift electricity demand to the moments when clean energy is most abundant.
With these tools, everyday appliances such as heat pumps, air conditioners, radiators, water heaters, EVs, and even household batteries can operate like virtual power plants. They can ease pressure on the grid during peak demand and step in when renewable production dips, reducing the need for costly and polluting backup pow
As we move into the age of electrification, AI’s role in energy management will be both vast and vital. Yes, it will drive additional electricity consumption. But it also provides the intelligence and responsiveness the energy transition depends on.
Without it, we may struggle to deliver on decarbonisation promises. With it, we have a real chance of making them a reality.
