Northern Virginia has become the epicentre of the global data centre boom, with AI driving explosive demand and stretching the limits of power systems. This article explores how utilities, regulators, and hyperscalers are balancing physics, politics, and planning to meet soaring demand, deliver clean power, and build resilience for the AI era.
Dominion Energy and other regional electricity providers in the region are rising to the challenge and blending infrastructure and AI inside the grid itself
Artificial intelligence (AI) data centres today are vast symbols of a new frontier in computing, with Northern Virginia in the US emerging as the world’s largest data centre market. To put it into perspective, the region is almost five times the size of any other market in the US. At the end of the first quarter of 2025, Northern Virginia’s data centre inventory topped 4,900 megawatts (MW), according to JLL’s quarterly report on global data centres.
The quest for super-intelligence is spurring a data centre boom as Big Tech pours in billions to expand the capacity and technology used to train their large-language models. But while this wave of AI-driven hyperscale investment is building computing capacity for technology companies, it is also putting a strain on the state’s grid, planning systems and climate targets simultaneously.
Virginia’s response is to blend infrastructure upgrades, new rate structures and increasingly AI-integrated grid operations, offering a strategic model for other regions.
The challenge: physics, politics and planning
The data centre queue for Dominion Energy, the primary electricity provider in Northern Virginia, has exploded. According to the company, as of December 2024, it had around 40 gigawatts (GW) in various stages of contracting, up 88% from July 2024. This surge has necessitated significant capital expenditure and a long-term transmission build-out with regional transmission organisation PJM, and neighbouring utilities. Projections suggest around 6.3% annual load growth in Dominion’s zone, potentially doubling load by 2039, marking a stark shift from years of flat demand.
Land use and local consent: The Virginian counties of Loudoun and Prince William exemplify the balancing act. On one hand, they benefit from data centre tax revenues and jobs; on the other, the counties grapple with concerns about noise, skyline encroachment and high-voltage infrastructure. State analysts estimate that Northern Virginia accounts for a double-digit share of global operational capacity, a concentration intensifying both advantages and backlash.
Under Virginia’s Clean Economy Act, Dominion must deliver 100% zero-carbon electricity by 2045, even amid rising loads. That forces a tight balance between near-term reliability (gas and grid upgrades) and long-term decarbonisation (offshore wind, solar, storage).
Under Virginia’s Clean Economy Act, the state's Dominion Energy must deliver 100% zero-carbon electricity by 2045, even amid rising loads.
Building smarter supply and smarter demand
- AI inside the grid: Dominion is employing predictive analytics and outage modelling (including a collaboration with the University of Connecticut) and pursuing grid modernisation, which includes advanced metering, hardening and automation to enable a more dynamic, AI-assisted grid. Meanwhile, PJM is deploying AI to speed up interconnection studies, which is a key bottleneck for integrating new generation and storage. A Google-led initiative is already automating tasks that slow queue processing, moving towards a real-time “digital map” of the network, and accelerating clean-capacity interconnection.
- Flexible compute as a feature: AI training workloads are more delay-tolerant than latency-sensitive services. This allows parts of hyperscale demand to shift across hours or days to better align with grid conditions, facilitating demand response, price-based curtailment and future data centre-as-grid-asset programmes.
- Firming the clean build: Dominion’s 2.6 GW Coastal Virginia Offshore Wind project, roughly 50% complete and expected by late 2026, anchors the zero-carbon supply build, alongside utility-scale solar batteries and selective thermal capacity.
How the state is managing the wave
Rate design for very large loads: Dominion is proposing bespoke rates for large customers to reflect connection and reliability costs and encourage flexibility. They are introducing more granular time-of-use and locational tariffs.
Transmission first: Dominion and PJM have mapped a multi-year transmission expansion, including new lines, substations and rebuilds, to relieve constraints and channel power to data centre clusters. Hardware remains critical, even as software plays a growing role.
Diversifying geography: With Data Centre Alley nearing saturation, developers are expanding along the I-95 corridor. Richmond is now the fastest-growing sub-market in Virginia, driven by grid headroom and planning speed.
What hyperscalers are doing differently
Rack-level efficiency: AI clusters are increasing rack densities from 30–50 kilowatts (kW) to triple-digits. Operators are adopting liquid cooling, optimising airflow and using DC distribution to cut losses -each percentage point saved relieves grid strain.
Shifting work to chase power: AI workloads are scheduled off-peak or across regions where permitted, thus smoothing peaks and enabling tariff-based incentives for predictable load-shaping.
Projected energy demand of US data centres by 2030
Estimated spend on data centres in the US between 2025 and 2030
The bottom line
Virginia is betting that intelligence plus infrastructure can reconcile hyperscaler demand with clean power mandates, and retaining hyperscale growth while honouring climate goals. Over 23GW of new capacity remain in the pipeline. The winners of the AI era will be those who build capacity fast, steer demand smartly and digitise their grid to keep pace.
ADIPEC 2025 is a global platform designed to address this dual imperative – building resilience in today’s systems and scaling intelligent solutions to enable the system-wide transformation required to accelerate inclusive global progress. To meet this moment, we must secure supply amid uncertainty while seizing the opportunities that energy unlocks.