Datacenter

At ISS AI Hosting, we deliver cutting-edge supercomputing power through our modular campus ecosystem, where customers can host a range of GPUs and ISS provides all of the accommodations across 14 hives for unparalleled efficiency and scalability. This configuration supports approximately 143,360 GPUs, powering up to 200 MW of IT load with a hybrid cooling approach—75% direct-to-chip (DTC) for optimal thermal management and 25% air cooling for flexible deployment—enabling seamless AI model training, inference, and multimodal development. Our portable pod-based architecture, housed in standard shipping containers, ensures rapid setup in strategic locations while integrating sustainable energy sources, making us the ideal partner for frontier AI projects demanding exascale performance and reliability.

The project plans to utilize the six existing Carrizo Aquifer wells as the primary water source, with current capacity believed sufficient for operational needs. To enhance sustainability, adopt direct-to-chip (DTC) cooling to save 15.5 MW and reduce water use by 20-30% compared to traditional evaporative systems; use recycled/gray water and monitor aquifer levels.

The design will support a 200 MW IT load through hybrid power (206 MW solar, 362 MW CHP with multi-stage waste heat recovery, 50 MWh batteries, and 5-7 day fuel storage), redundant fiber connectivity to San Antonio and Laredo NAPs, and a hybrid cooling system (75% direct-tochip liquid, 25% air, both N+1 redundant). Beyond these, engineers will address structural resilience, fire suppression, security, IT redundancy, environmental controls, and operational sustainability to holistically secure Tier IV status.