Artificial intelligence is driving rapid growth in electricity demand, but data science may also play a central role in helping power systems cope. As AI data centres, electric vehicles and electrified industry place heavier and less predictable demands on the grid, forecasting, optimisation and real-time analytics are becoming increasingly important.
The challenge is no longer simply whether enough electricity can be generated over time. Grid operators must also determine whether power can be delivered to the right location at the right moment, particularly as new digital infrastructure develops faster than transmission and distribution networks can be expanded.
AI is changing the shape of electricity demand
According to the International Energy Agency, global electricity demand increased by 4.3% in 2024, almost twice the average annual growth rate recorded during the previous decade. Demand is expected to continue rising as data centres, transport and industrial processes rely more heavily on electricity.
AI infrastructure is placing particular pressure on electricity networks. Traditional high-performance computing facilities were generally based on CPU systems, lower rack densities and air cooling, allowing operators to expand capacity gradually.
GPU-intensive AI systems require much greater power density. They may also depend on liquid cooling, redesigned electrical systems and significantly larger grid connections. New data centre campuses are increasingly being planned at scales of hundreds of megawatts, rather than the tens of megawatts associated with earlier facilities.
This creates a complex planning problem. Utilities and developers need to forecast future loads, assess where grid constraints are likely to emerge and determine how new demand can be accommodated without reducing reliability elsewhere.
Europe faces growing grid constraints
The challenge is especially pressing in Europe, where much of the electricity network was built before the rapid growth of renewable generation, electric vehicles and large digital loads.
Electricity demand across the European Union rose by around 1.5% in 2024, reversing several years of limited growth. At the same time, renewable energy projects are facing lengthy connection queues, while shortages of transformers, high-voltage cables and specialist workers are slowing the delivery of new infrastructure.
Grid upgrades can take many years to plan and complete. Data centres and other major electricity users, however, often want to begin operating much sooner. This mismatch is increasing interest in technologies that can make better use of existing network capacity.
Battery storage as a flexible bridge
Battery energy storage is emerging as one potential solution. More than 42GW of battery storage capacity was added globally in 2023, making it one of the fastest-growing commercially deployed energy technologies.
Batteries do not remove the need for new grid infrastructure. Instead, they can provide flexibility while longer-term upgrades are planned and delivered.
A battery system can draw electricity gradually when demand is lower and release it during peak periods. This can help reduce local congestion, smooth sudden changes in demand and allow projects such as data centres, renewable energy developments and electric vehicle charging hubs to make better use of an existing connection.
The falling cost and relatively short construction time of battery systems also make them attractive. While major transmission projects can take years, storage facilities may be developed in months.
A growing role for data science
Effective battery deployment depends on more than installing hardware. Operators need accurate information about electricity demand, weather, renewable generation, network conditions and energy prices.
Machine learning and statistical forecasting can help predict when demand will rise or renewable output will fall. Optimisation models can then determine when batteries should charge, when they should discharge and how available capacity should be allocated.
Real-time analytics can also support faster responses to changing conditions, while digital twins may allow planners to test different combinations of storage, grid reinforcement and flexible demand before committing to expensive physical investments.
These tools become increasingly valuable as electricity systems grow more complex. Instead of managing relatively predictable flows from large power stations to consumers, grid operators must now coordinate renewable generation, distributed storage, electric vehicles and major new digital loads.
AI creates the problem and helps shape the solution
The rapid expansion of AI infrastructure is exposing the limits of existing electricity networks. Yet many of the same data science techniques supporting AI development may also help utilities and developers manage its impact.
Better forecasting can identify emerging constraints earlier. Optimisation can make more efficient use of batteries and existing grid connections. Real-time data can help operators respond as demand changes.
Battery storage is therefore becoming more than an energy technology. Combined with data science, it could provide an important bridge between a digital economy growing at speed and an electricity system that will take much longer to transform.
References
https://www.energy-storage.news/in-the-age-of-ai-data-centres-and-electrification-the-grid-is-the-real-bottleneck/
https://www.iea.org/reports/global-energy-review-2025/electricity
https://www.weforum.org/stories/artificial-intelligence/electricity-data-grid-connectivity-strategic-bottleneck-ai-transformation/