Artificial intelligence has arrived in the energy transition in two very different ways.
On one hand, AI could help us run cleaner, more flexible and more efficient energy systems. It can forecast demand, optimise networks, improve maintenance, support trading, accelerate engineering design and help consumers use energy more intelligently.
On the other hand, AI is creating a very large new source of electricity demand.
That makes the question more interesting than it first appears. AI may accelerate the energy transition, but only if the power needed to run it does not slow the transition down.
The demand shock
For years, electricity demand in many mature economies was relatively flat. Efficiency gains, deindustrialisation and improvements in appliances helped offset new sources of demand.
That picture is now changing.
Electrification was already expected to increase electricity consumption as cars, heating and industry shift away from fossil fuels. AI has added another layer. Data centres are no longer just background digital infrastructure. They are becoming major energy assets.
The International Energy Agency estimates that data centres used around 415 TWh of electricity globally in 2024, around 1.5% of global electricity consumption. In its base case, this could more than double to around 945 TWh by 2030. That is still a relatively small share of global electricity demand, but the pace of growth is striking.
The challenge is not only the total amount of electricity. It is where and how the demand appears.
A data centre can be built much faster than a new transmission line, offshore wind farm or nuclear plant. It may also be concentrated in specific locations, often close to fibre networks, major cities, financial centres or existing cloud infrastructure. That can create pressure on local grids long before the national energy system appears short of power.
In Great Britain, NESO has highlighted how quickly the picture is evolving. Its analysis points to around 5.2 GW of connected data centre capacity and just over 20 TWh of annual electricity demand by 2030. However, connection requests are far higher, with more than 70 GW in the pipeline. Not all of this will be built, but the signal is clear: AI and cloud growth are becoming part of energy system planning.
The same issue is visible in the United States. The Department of Energy has pointed to a return to rising electricity demand, driven by AI, data centres, manufacturing and wider electrification. Some estimates suggest data centres could rise from around 4% of US electricity load in 2023 to as much as 9% by 2030.
This is a remarkable change.
The energy sector is used to planning around power stations, networks, homes and factories. It must now also plan around compute.
How AI could help utilities
For utilities, AI offers practical opportunities across the value chain.
Electricity networks are becoming more complex. Power is no longer only flowing from large central power stations to passive consumers. Increasingly, electricity flows in both directions, with rooftop solar, batteries, electric vehicles, heat pumps and flexible demand all interacting with the system.
AI can help manage this complexity.
It can improve forecasting of wind, solar and demand. It can help identify faults before they occur. It can optimise network reinforcement decisions. It can support control rooms by identifying patterns across millions of data points. It can also help utilities engage with customers more effectively, from smarter tariffs to better advice on energy use.
For generators, AI can improve asset performance. Wind farms, solar plants, batteries and thermal assets all produce vast amounts of operational data. Better analytics can improve maintenance, increase availability and reduce costs.
For system operators, AI may become essential. A cleaner electricity system will be more weather-dependent, more decentralised and more dynamic. Human operators will still be vital, but they will need better digital tools to manage the scale and speed of change.
What it means for investors
For investors, AI creates both opportunity and risk.
The opportunity is clear. More electricity demand means more investment in generation, storage, networks, flexibility and grid infrastructure. Data centres also create new demand for clean power purchase agreements, behind-the-meter energy solutions, batteries, backup generation and heat reuse.
This could accelerate investment in renewable energy and flexible assets.
However, there are risks. Data centre demand forecasts are uncertain. Connection queues may overstate what is genuinely viable. Some projects may never be built. Others may be delayed by grid constraints, planning issues, supply chain bottlenecks or public concern over water and land use.
Investors therefore need to distinguish between speculative demand and bankable demand.
There is also a strategic question. Should new data centres be treated simply as large electricity consumers, or should they be designed as active participants in the energy system?
The best projects may be those that combine location, clean power, flexibility, heat recovery and grid support. A data centre in the wrong place can create system costs. A data centre in the right place, with the right commercial structure, could help absorb renewable generation and support local energy systems.
What it means for consumers
For consumers, AI may feel distant. Most people do not see a data centre when they ask a chatbot a question, stream a film or store photos in the cloud.
But the impact may still reach households.
If electricity demand grows quickly and infrastructure does not keep up, costs can rise. New generation, grid connections and network reinforcement all need to be paid for. Poorly planned growth could increase pressure on bills.
At the same time, AI could help consumers benefit from the energy transition.
Smarter systems could help households decide when to charge an electric vehicle, when to use a battery, when to export solar power and how to reduce energy use at peak times. AI could make energy tariffs easier to understand and help people use energy in ways that are cheaper and lower carbon.
The risk is that the benefits are unevenly distributed. Consumers with solar panels, batteries, EVs and smart appliances may be able to participate actively. Others may have less flexibility and fewer options.
A fair transition will need to ensure that AI-enabled energy systems do not only benefit the most digitally connected or wealthier households.
The challenges
AI is not a magic solution.
It brings real challenges for the energy sector.
The first is electricity demand. AI data centres need reliable, high-quality power. If that demand is met by fossil generation, the emissions impact could be significant.
The second is grid capacity. Many electricity networks were not designed for rapid, concentrated demand growth. Connections, substations and transmission lines can take years to plan and build.
The third is resilience. More digital control can improve performance, but it also increases the importance of cybersecurity and operational discipline. Critical infrastructure must be intelligent, but it must also be secure.
The fourth is transparency. Energy planners need better data on where data centres are likely to connect, how much electricity they will actually use, how flexible their demand can be and whether they will provide system benefits.
The fifth is governance. AI will increasingly influence decisions about infrastructure, markets and consumers. The energy sector will need clear rules on accountability, data quality, explainability and trust.
So, can AI accelerate the energy transition?
Yes, but not automatically.
AI can help design, operate and optimise a cleaner energy system. It can make renewable assets more productive, networks more efficient and consumers more empowered. It can help utilities make better decisions and help investors find new opportunities.
But AI also increases the scale and urgency of the energy transition by adding a major new source of electricity demand.
The answer depends on how we plan.
If AI data centres are connected without strategic coordination, they could add costs, increase emissions and create new constraints. If they are planned alongside clean generation, storage, flexible demand and network investment, they could become part of the solution.
The energy transition was already about electrifying the economy.
AI reminds us that the digital economy is also physical. It needs land, water, wires, power stations, substations and people.
The most important question may not be whether AI can accelerate the energy transition.
It is whether we can build an energy system capable of supporting AI while still delivering secure, affordable and low-carbon power for everyone.
