How Data Centers Could Become Part of the Power Grid
The AI boom is creating a new challenge for the energy system: how to power an unprecedented wave of data centres without putting even more pressure on already-constrained grids.
The International Energy Agency expects global electricity demand to grow by an average 3.6% a year between 2026 and 2030, driven partly by data centres, industry, electric vehicles and cooling. Meanwhile, more than 2,500 GW of generation, storage and large-load projects are waiting in grid connection queues worldwide.
But data centres may not always have to be treated simply as another source of demand.
Not every computing task needs to happen immediately. AI training, batch processing and other flexible workloads can potentially be delayed, accelerated or moved to another location depending on electricity availability and grid conditions.
That creates an emerging idea: data centres could adjust their electricity consumption in response to the grid, much like other flexible energy resources.
Research published in 2026 found that around 20% of data-centre power consumption in the study’s real-world workloads came from tasks suitable for load shifting. The researchers demonstrated that these workloads could provide meaningful demand flexibility during periods of grid stress.
Instead of asking only “How much electricity will this data centre consume?”, utilities could increasingly ask: “How flexible is that demand?”
Why It Matters
The opportunity becomes particularly interesting as renewable energy expands. Solar and wind generation can produce large amounts of electricity when demand is low and less when demand is high.
Flexible computing could help bridge that mismatch. AI workloads could potentially be scheduled for periods when renewable electricity is abundant, while non-urgent workloads could be shifted away from periods of peak demand.
The International Energy Agency increasingly identifies demand-side flexibility as an important part of managing growing electricity demand and variable renewable generation.
The Bigger Picture
This doesn’t mean data centres will suddenly become power plants. Critical computing workloads still need reliable, uninterrupted electricity, and not every task can be moved or delayed.
But as AI makes data centres one of the fastest-growing sources of electricity demand, flexibility could become just as important as efficiency.
The next generation of data centres may therefore be designed not only to consume electricity efficiently, but to respond intelligently to when, where and how that electricity is available.
That could turn one of the biggest new pressures on the grid into a potential source of flexibility.
References
- International Energy Agency (IEA). Electricity 2026: global electricity demand growth, grid connection bottlenecks and the growing impact of data centres.
- International Energy Agency (IEA). Energy and AI: the energy implications of AI and data centres, including opportunities for flexible computing demand.
- International Energy Agency (IEA). Flexibility, Electricity 2026: the role of demand-side flexibility in managing growing electricity demand and variable renewable generation.
- Caprara et al. (2026). Data Center Workload Flexibility for Power System Demand Response: research using real-world Alibaba workload traces to explore how flexible computing workloads could support grid demand response.
Get innovation insights from The FutureList weekly. Subscribe to our newsletter here