Galaxy Digital’s 500-Acre Texas AI Campus Signals a New Grid Model for Large Loads

Galaxy Digital’s latest AI data center project in McGregor, Texas, illustrates a fundamental shift in how utilities and regulators approach massive power demands from artificial intelligence workloads. Developers are increasingly expected to fund grid upgrades upfront and demonstrate that their load commitments are real before utilities invest billions in new transmission infrastructure.

The company announced Tuesday it has acquired 500 acres in McGregor, about 90 miles north of Austin, where it plans to build an AI and high-performance computing campus. The project, internally known as “Project Merlin” during city council discussions, will start with a 74 MW first phase and scale into a multi-hundred-megawatt campus by 2030 as additional transmission capacity becomes available. Galaxy said the first phase is expected to begin receiving power in 2028.

This is Galaxy’s second major Texas AI campus, following its 1.6 GW Helios development in Dickens County. But unlike many large-scale announcements, Galaxy has emphasized a set of self-funded commitments: it will privately finance an on-site electrical substation, provide financial assurances for utility upgrades, and cover additional water infrastructure required under its development agreement with the City of McGregor. The total planned investment exceeds $400 million, and the campus is expected to create at least 30 full-time jobs with average annual salaries above $60,000. It will include six to eight data center buildings, with water consumption capped at approximately 3,000 gallons per day per powered shell and noise limited to 65 dBA at the property boundary using a multi-layered “box within a box” design with 11-inch precast concrete walls.

The project aligns with recent policy changes in Texas. Last month, Governor Greg Abbott directed the Public Utility Commission and ERCOT to develop rules requiring data center developers to pay for required electric infrastructure while strengthening oversight of water use and other impacts. According to Galaxy’s presentation to the McGregor City Council, many of the campus commitments—including the private substation, closed-loop cooling, contractual water caps, and noise limits—had already been negotiated before Abbott issued the directive.

“The emerging model is moving from ‘beneficiary pays’ to ‘beneficiary proves,’” said Neil Osnato, founder of Persistence Analytics Group. “Developers are increasingly being asked not only to fund the infrastructure they cause, but to prove that the underlying load is durable enough to justify the grid being planned around it.” Osnato noted that paying for the substation solves cost causation but not demand verification. Utilities are placing greater weight on whether proposed AI loads are backed by committed customers, financing, site control, permits, and construction milestones before making long-term grid investments.

Galaxy enters the McGregor market with a track record. Earlier this month, the company said its Helios campus had delivered roughly 200 MW of gross power, including 133 MW of critical IT load, under a long-term lease with CoreWeave. The McGregor campus will add at least $130 million to the city’s property tax base, generate about $7.5 million in land-sale revenue, create several hundred construction jobs, and operate without a property tax abatement. “We’ve spent the last several years proving in Dickens County that a privately funded data center can be a real, long-term partner to a rural community,” said Galaxy Founder and CEO Mike Novogratz. “McGregor is the next step in our strategy to build a disciplined, multi-campus data center business.” Under the development agreement, Galaxy will finance and build its own private substation and provide financial security for utility upgrades, with the second phase positioning the company as an “anchor tenant” to support future transmission expansion that could accommodate additional industrial development in the region.

The McGregor campus underscores a broader trend: as AI developments grow into the hundreds of megawatts, developers will compete not only for power but also for the confidence of utilities willing to build around them. Projects that can prove both financial commitment and operational durability are likely to gain preference in an increasingly constrained grid environment.