Army Partners with Private Firms to Build Data Centers

Truthout reports that in late March 2026 the U.S. Army selected private companies to build and operate two hyperscaled data centers, sited at Fort Bliss, Texas, and Dugway Proving Ground, Utah. According to an Army spokesperson quoted by Truthout, the Army has entered an "exclusive negotiation period" with the companies to negotiate "specific lease economics" and the proposed leases are "long term, 50-year" arrangements. The spokesperson told Truthout that the Army will accept "in-kind consideration," meaning it will receive services or improvements in lieu of cash rent, and confirmed the projects "are indeed commercial data centers" that may sell surplus computing capacity commercially.

Editorial analysis - technical context: The Truthout piece does not list the developer names, specific hardware, or on-premises architectures. Public reporting of similar hyperscale projects typically centers on investor-backed, commercially operated facilities that provision dense GPU clusters and bespoke networking for high-throughput AI workloads. For practitioners, the most relevant technical unknowns are compute density (GPUs per rack), cooling and power arrangements, connectivity to classified networks, and whether the facilities will offer dedicated secure enclaves versus colocated commercial capacity.

Truthout frames the deals as part of two converging trends: heavy Wall Street financing of data centers and the U.S. military's growing use of AI. Public-works leases that exchange long-term facility access or capability for in-kind services can broaden available compute for defense applications while also creating commercial supply in the same sites. For the broader compute market, partnerships that free military demand into privately financed builds can affect capacity planning, regional colocation prices, and the pool of commercially available AI compute capacity.

Editorial analysis: Observers should track which developers and financiers are announced, the contractual perimeter for classified versus unclassified workloads, and any future disclosures about installed GPU/accelerator types and networking. Also monitor whether lease terms, when finalized, include guaranteed dedicated compute allocations, facility-level air-gap provisions, or resale restrictions on commercial capacity. These details determine how useful the facilities will be for large-scale model training, inference at the edge, or classified experimentation.

Truthout is the sole published source in this packet and the Army spokesperson quoted by Truthout provided the key details cited above. The developer identities, exact financing partners, and specific technical specifications were not disclosed in the Truthout article.