In September 2026 the International Energy Agency (IEA) published Modernising Grids in the Age of Electricity. The report estimates that using existing grids more efficiently could allow up to 330 GW of generation, storage and demand to connect without reinforcing the network.
For data centre grid connections, transformer shortages and connection queues have been the biggest constraint. The report matters because it quantifies how much can be done to draw capacity out of existing assets, rather than simply waiting for new lines to be completed.
330 GW Without Reinforcement, About USD 100 Billion With It
The estimate covers the three technologies most widely deployed on transmission grids.
The three technologies in the IEA estimate
01
Dynamic ratings
Adjust how much power can safely flow through lines and transformers according to actual weather conditions.
02
Topology optimisation
Reconfigure the network to route power around congested lines.
03
Advanced power-flow control
Redirect power onto under-used paths.
According to the IEA, most of these can be deployed in months rather than years and retrofitted onto networks already in operation. However, the capacity they release is limited by the physical network layout and by weather and system conditions, and they do not deliver the step changes in capability needed for demand growth. The IEA positions them as a complement to, not a substitute for, grid reinforcement.
Results are already appearing on real systems. In the United States, dynamic ratings have avoided USD 64 million a year in congestion costs, and in Great Britain advanced power-flow control has added more than 2 GW of transmission capacity. In one case, a detailed virtual model of the grid (a digital twin) cut the time needed to analyse reinforcement options by 70%.
1,700 GW of Renewables and 600 GW of Batteries in the Queue
Behind this is the reality that grid build-out has not kept pace with demand growth. According to the IEA, at least 1,700 GW of advanced-stage renewable projects and 600 GW of grid-scale batteries were waiting to connect as of 2025. Grid congestion costs in 2024 reached USD 12 billion in the United States and EUR 4.3 billion in the EU.
Over the past decade, electricity demand has grown about twice as fast as overall energy demand, driven by electrification of end uses, cooling, industrial expansion and, in some markets, data centres. Meeting demand growth to 2035 will require grid capacity to expand by at least 30%, with 25 million km of lines built or refurbished.
In the report, the IEA describes data centres as among the most spatially concentrated grid users, because they cluster near cities, fibre networks and customers. It notes that in areas where the grid cannot keep up, housing and industrial development compete with data centres and generation projects for scarce grid capacity. The picture overlaps with that shown by the US transmission needs study.
The Barrier Is Institutions, Not Technology
Loading chart
Network operators surveyed by the IEA cited skills and organisational readiness (64%), data availability and quality (60%), trust-related issues (60%) and regulatory frameworks (56%) as the main barriers. The IEA concludes that the main obstacles to deployment are institutional and organisational rather than technical performance.
On the regulatory side, operators often still find it easier to recover spending on new equipment than on digital or operational alternatives that solve the same problem. The IEA recommends funding rules that do not favour capital expenditure, and incentives tied to outcomes such as capacity released or congestion avoided.
Attitudes towards AI also differ by use. While 70% of network operators use AI widely in maintenance, only 23% use it in real-time operations, because requirements for explainability, auditability and accountability weigh heavily on safety-critical decisions.
What It Means for Data Centre Developers
How to frame connection talks: Beyond asking the utility when the new substation will be ready, there is room to ask how much capacity could be released earlier through better operation of existing assets
Understanding the limits: Released capacity varies with weather and system conditions, so it cannot be applied as-is to designs that assume a constant supply at all times
