As AI drives massive demand for data center capacity, the industry is entering a more complicated phase of growth. Three recent developments illustrate the changing landscape: the collapse of an ambitious data center project, new state legislation drafted to protect electricity customers from data center costs, and yet another multibillion-dollar commitment from an AI company seeking long-term compute capacity.

Together, the three stories show that while demand for AI infrastructure remains exceptionally strong, developers face stricter regulation, greater community scrutiny and more complex economics before bringing new capacity online.

Enormous Data Center Project Cancelled

A clear sign of a shift came in Virginia, where the long-running Prince William Digital Gateway proposal has officially come to an end.

QTS Realty Trust withdrew its appeal to the Virginia Supreme Court, ending a legal battle surrounding what would have been a 2,100-acre campus with 37 data center buildings totaling roughly 22 million square feet. The project represented an estimated value of approximately $100 billion and, if completed, would have ranked among the largest data center developments ever proposed.

The project unraveled after Virginia’s Court of Appeals invalidated the county’s rezoning approval because required public notice procedures were not properly followed. Compass Datacenters had already abandoned its own appeal, and Prince William County also elected not to continue litigation. The proposal became a focal point for local opposition over land use, power demand, water consumption and its proximity to Manassas National Battlefield Park.

Most significant about the project’s collapse: it exemplifies the resistance facing large developments. Communities across the country are paying closer attention to the impact data centers have on electric grids and surrounding neighborhoods, forcing developers to devote more attention to permitting and public engagement.

New Jersey Approves Legislation to Protect Ratepayers

State governments are also reshaping how data centers connect to the power grid. New Jersey lawmakers approved legislation directing the state’s Board of Public Utilities to establish a dedicated electricity tariff for data centers and other facilities requiring at least 50 megawatts of capacity. The measure aims to prevent residential and small business customers from absorbing infrastructure costs created by large new loads.

The proposal also requires qualifying facilities to guarantee payment for at least 85 percent of their requested electric service for 10 years, even if they ultimately consume less power. Multiple facilities under common ownership or located on adjoining properties would be treated as a single project when determining whether they exceed the 50 MW threshold.

Developers would also need to demonstrate that proposed projects are genuine rather than speculative, while facilities making binding commitments to provide their own clean generation or energy storage would receive priority during the interconnection process. During grid emergencies, large data centers would also face curtailment before residential customers.

New Jersey joins a growing list of states introducing policies intended to balance rapid AI-driven infrastructure expansion with grid reliability and consumer protection.

TeraWulf Inks 20-year Agreement with Anthropic

Despite these additional hurdles, AI companies continue to invest heavily to lock in enormous amounts of future computing capacity.

TeraWulf announced a 20-year agreement with Anthropic for its Justified Data campus in Hawesville, Kentucky. The campus is expected to provide approximately 401 megawatts of critical IT capacity, with initial power scheduled for the second half of 2027 and full build-out anticipated by early 2028.

The agreement is projected to generate roughly $19 billion in contracted revenue over the initial lease term, providing TeraWulf with a long-duration revenue stream while further demonstrating that leading AI developers are eager to secure dedicated infrastructure years before facilities become operational.