top of page

Q2 2025 Data Center Activity Report


Q2 2025 Data Center Activity Report


The U.S. data center sector accelerated in Q2 2025 as hyperscale operators ramped up AI training and inference workloads. Total energy consumption for the quarter reached an estimated 224TWh driven by new GPU dense pods and edge expansions. Capital expenditures reached a growth rate of 17.7% in installed capacity with hyperscalers accounting for over $300 billion in capital expenditures. Despite ongoing challenges which include workforce shortages and interconnection delays, developers found opportunity in new state-level incentives and deregulation tied to the “Big Beautiful Bill,” which helped unlock rural builds across key markets. Regional winners this quarter included Ohio, Georgia, and Texas. This report dives into those market dynamics, key technology trends, energy and land shifts, policy impacts, and a deeper look at Ohio’s fast-growing corridor.


US Data Center Development Q2 2025
Data Center Count Q2 2025. Source: LandGate

Evolving Data Center Energy Strategies


Hyperscalers are rapidly diversifying their energy stacks to ensure reliability, cost control, and alignment with carbon reduction goals. In Q2, a wave of new energy strategies emerged across major campuses, including mixed-source microgrids that blend on-site solar, small modular reactors (SMRs), and gas turbine peakers. These configurations are designed to meet the unpredictable and high density demands of AI computing. Increasingly, data centers are turning to private wire arrangements that connect directly to independent power producers, bypassing traditional utilities and avoiding lengthy interconnection queues. Behind-the-meter solar and storage arrays are also gaining traction, especially as operators pursue 24/7 carbon-free energy compliance. Notably, private wire deals in Indiana and Ohio set new benchmarks this quarter in both pricing and contract length, reflecting a broader shift toward long-term power autonomy. Still, the task of matching volatile AI load profiles particularly during training and inference surges with intermittent solar generation remains one of the most pressing engineering challenges facing the industry.





Want to read more? Access the full report below:




bottom of page