The AI race is usually framed as a contest for chips, models and talent. Increasingly, it is being decided by something more basic: power.
The International Energy Agency (IEA) projects that electricity demand from data centers will double by 2030.
At the same time, many grids are already under strain from increased electrification, ageing infrastructure and long permitting cycles.
Latest Videos FromTechRadar
EVP Power Systems at Schneider Electric.
This makes the ability to secure reliable electricity on a timeline that matches business plans a decisive factor in choosing where to build an AI data center.
Time to power is becoming a key constraint
That is why “time to power” is quickly becoming the key metric in site selection.
Schneider Electric’s work in this space shows that in multiple regions, grid connection approvals and reinforcement timelines that were once manageable now stretch far beyond normal development cycles.
In energy-constrained hotspots, the process to connect to the grid can extend for years, and in some cases can reach a decade or more. For large AI data center projects, that changes where data center operators build and how they build.
The issue is not the lack of power. The real challenge is that data centers concentrate in specific regions, and those local grids hit capacity limits faster than new infrastructure can be planned, approved and built. When that happens, power availability starts to overtake “location” as the deciding factor.
Transparency is key to attracting investment and ensuring resilience
Power availability also influences where companies are investing when it comes to building AI data centers as it is key to resilience. The most attractive markets are those that offer certainty and transparent information on available grid capacity and reinforcement plans.
Just as importantly, data center operators need confidence that grid connection queues are managed in a way that prioritizes projects that are ready to build, rather than allowing speculative projects to sit in line for years and block capacity.
Resilience has always been fundamental to data centers, but it’s now being redefined. It is no longer only about redundancy inside the facility; it is equally about resilience to the grid environment around it. In many energy-constrained regions, shrinking grid capacity and congestion increase the risk of outages and emergency grid measures.
Hybrid power generation strategies offer an alternative to legacy grid dependence
These pressures are driving many data center developers and owners to consider on-site hybrid energy strategies which combine on-site power generation from renewables with electricity provided by the grid.
Such hybrid power approaches are key for reducing reliance on legacy grid infrastructure and enabling data center operators to generate energy on-site and improve resilience. This also increases energy flexibility by allowing data center operators to change when and how energy is used, stored, or generated in response to demand, prices, or grid conditions.
A concept that is getting more attention in this space is “energy parks”. In simple terms, an energy park brings together multiple power resources, from renewables, thermal power, and storage to on-site generation, and connects them to the grid through a single point. These energy parks may connect to the grid through a single interconnection point, operate as fully islanded systems, or combine both approaches through hybrid architectures.
Because the generation, storage and grid connection are designed together, energy parks can offer a faster or more predictable “time to power” in places where the traditional grid connection queue is a challenge. This could help alleviate the grid connection issue for many data center developers.
A new approach to designing data centers
This is where energy strategy becomes a design strategy. Developers need to plan across three key areas: how to connect, how to operate, and how to scale.
Connecting to the grid is now about getting the earliest possible clarity and avoiding prolonged delays. Operating is about building for an environment where grid constraints, curtailment risk and price shocks can impact business continuity. Scaling is about making sure a site that works at 50 MW does not become constrained at 150 MW because the upstream network cannot grow with the campus.
But none of this can be separated from sustainability. Sustainable AI infrastructure at scale requires electrification strategy and grid modernization: faster grid reinforcement, better planning, and more flexibility on both the supply and demand sides. It also requires closer coordination between developers, grid operators, policymakers and clean energy providers. This will help ensure that digital infrastructure growth is aligned with clean capacity additions and realistic delivery timelines.
For data center developers, the practical takeaway is straightforward: start with power and treat energy as a critical design variable. Build a site selection process that tests grid readiness and timelines before land is locked in, and design an energy strategy that covers connection, resilience and scale from day one.
When it comes to attracting AI infrastructure investment, the winners will be those that can offer speed and certainty on grid connection and reinforcement, and enable credible pathways to clean, resilient power.
Best web hosting services: 60 sites tested to find the 10 fastest, most reliable platforms.
This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.
The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit
