TeraWulf and Fluidstack priced project-level financing for their 168 MW high-performance computing (HPC) joint venture at the Abernathy, Texas campus, advancing development of a large-scale, liquid-cooled AI data center. The facility targets up to 240 MW of gross power capacity, with 168 MW allocated to critical IT load, under a long-term hosting structure backed by investment-grade credit support.
The financing structure draws on long-term credit enhancement delivered through Fluidstack’s platform and supported by a global hyperscale partner. TeraWulf plans to use the proceeds to fund construction, establish reserves, and complete delivery of the site. The companies expect commissioning in the second half of 2026.
The Abernathy project extends a broader partnership strategy focused on repeatable site design, disciplined capital formation, and rapid deployment of AI infrastructure. The joint venture also preserves optionality to expand beyond the initial 168 MW, using existing land, transmission access, and development work to support future high-density capacity as demand for HPC compute continues to rise.
- Project financing covers a 168 MW critical IT load, within a 240 MW gross power design
- Facility uses liquid cooling to support high-density AI workloads
- Long-term hosting structure includes investment-grade credit enhancement via Fluidstack
- Site remains on schedule for commissioning in 2H 2026
- Joint venture retains expansion potential beyond the initial phase
“This financing represents another important step in scaling a platform that was designed from the outset to grow,” said Paul Prager, Chief Executive Officer of TeraWulf. “We focused early on building sites, relationships, and structures that could adapt as the AI market evolved. Today’s environment validates that approach.”
🌐 Analysis
Project-level financing for AI colocation continues to mature as hyperscalers and AI platforms seek diversified, execution-ready capacity outside traditional cloud regions. TeraWulf’s approach mirrors a broader industry shift toward power-secured, liquid-cooled campuses that support rapid scale while improving financing terms through long-term credit structures.
TeraWulf Inc. is a U.S.-based digital infrastructure company headquartered in Easton, Maryland, focused on developing and operating low-cost, environmentally responsible data center infrastructure optimized for high-performance computing (HPC) and AI workloads. Founded in 2020 by co-founders Paul Prager and Beowulf Energy veterans, the company is led by Chairman and CEO Paul Prager, who has a background in energy infrastructure and power generation. TeraWulf’s core strategy centers on vertically integrated campuses powered predominantly by zero-carbon energy, including hydroelectric and nuclear sources, with flagship facilities such as the Lake Mariner campus in New York and the Nautilus Cryptomine facility in Pennsylvania. Originally known for sustainable Bitcoin mining, TeraWulf has increasingly repositioned its infrastructure to support AI and HPC use cases, including liquid-cooled data center designs and long-term hosting arrangements with hyperscale and AI platform partners. The company is publicly traded on Nasdaq (WULF) and has financed its growth through a mix of equity, asset-level project financing, and joint ventures, with recent milestones including large-scale HPC project financing and partnerships aimed at accelerating AI-ready capacity deployment.
Fluidstack is a cloud infrastructure company headquartered in London, focused on providing large-scale GPU compute for artificial intelligence and high-performance workloads by aggregating and orchestrating underutilized hardware across a global network of data centers. Founded in 2017 by CEO César Makary, the company’s mission is to deliver affordable, elastic, and energy-efficient AI compute by decoupling cloud services from the traditional hyperscale data center model. Fluidstack’s core platform uses software-defined orchestration to pool GPUs from enterprise, colocation, and partner facilities, enabling customers to access NVIDIA-based clusters for training and inference without long-term commitments. The company serves AI labs, enterprises, and cloud-native developers, and has increasingly positioned itself as an infrastructure partner for large AI data center and HPC projects. Fluidstack is venture-backed, with funding from investors including Speedinvest and LocalGlobe, and has expanded its footprint across Europe and North America through partnerships with data center operators and energy-aligned infrastructure providers, marking its evolution from a distributed GPU marketplace to a strategic AI infrastructure platform.
