ThisWay Global Acquires Bankrupt Site for 73 MW AI Hub
Texas AI software firm converts stalled RELLIS campus facility into liquid-cooled compute hub
ThisWay Global has acquired an unfinished, bankrupt data center project at Texas A&M’s RELLIS Campus in Bryan, Texas, to build a 73 MW AI compute center. The project converts a stalled colocation development into a high-density facility equipped with closed-loop liquid cooling for sovereign research and enterprise workloads.
Key details
Austin-based AI software company ThisWay Global (TWG) completed the acquisition of the unfinished data center out of Chapter 11 bankruptcy on September 30, 2026, making creditors whole in partnership with Centaurus Capital and Skylar Capital Management. The facility sits on 25 acres of Texas A&M University System (TAMUS) land where data center development had languished since 2013 before the original private developer entered bankruptcy in November 2025.
The planned Research and AI Compute Center (RACC) will scale to 73 MW across multiple phases. The initial building accounts for roughly 25% of the total campus capacity. While the original bankrupt design contemplated conventional air-cooled colocation and office space, TWG is re-engineering the structural and thermal architecture to deploy high-density high-performance computing (HPC) racks.
To support high-density AI training and inference, TWG is installing a closed-loop liquid cooling system utilizing a propylene glycol and water mixture. Power deployment from Bryan Texas Utilities (BTU) will proceed in phases tied to utility approvals, while TWG runs its Amalgamy.AI software stack to enforce sovereign access controls.
Why this matters
The acquisition demonstrates how stalled or distressed commercial data center real estate is being retrofitted specifically to meet the high power and thermal densities demanded by modern AI infrastructure. By replacing traditional air cooling with closed-loop liquid cooling, the facility aims to support high-density GPU racks while reducing external water consumption compared to evaporative cooling towers.
Furthermore, co-locating 73 MW of AI compute capacity on public university land highlights a growing trend of institutional partnerships aimed at providing secure, localized compute resources for regulated sectors like defense, healthcare, and energy.
Context
Data center developers in Texas and across the United States face increasing scrutiny over grid capacity and water usage. The RELLIS project's transition from a standard colocation facility into a phased liquid-cooled AI hub reflects broader industry shifts toward retrofitting existing structures rather than starting new greenfield developments. Closed-loop cooling architectures have become essential for operators aiming to bypass municipal water constraints while supporting GPU rack densities exceeding 100 kW.
What happens next
ThisWay Global will complete structural retrofits on the initial 25% compute block and begin installing liquid-cooled server racks. Phase deployment will depend on formal interconnection approvals and capacity allocations from Bryan Texas Utilities. Once operational, the center will serve research, defense, and healthcare partners while testing TWG's Sovereign AI Blueprint for verifiable data access.
Source: Data Center Knowledge Published on AI Usage Global, author: AUG Bot
