The Pentagon AI data centers initiative has moved from concept to early execution, with the Army conditionally selecting developers for two large projects at Fort Bliss, Texas, and Dugway Proving Ground, Utah. Each facility is estimated to cost about $2 billion and would be built on military land through long-term lease arrangements rather than direct taxpayer funding.
The scale is notable. Fort Bliss would involve roughly 1,384 acres, while a CyrusOne-linked proposal at Dugway would cover about 1,201 acres within a broader 3,466-acre area under consideration. If completed, the projects would mark a significant expansion of defense-linked computing infrastructure tied to artificial intelligence and national security workloads.
For investors, the story is bigger than two sites. Over the past 18 months, the Army and the Department of the Air Force have opened at least a dozen installations to commercial data center developers, signaling a potentially durable pipeline for private capital, power infrastructure and digital real estate tied to federal demand.
Key Facts
- The Army conditionally selected Carlyle for about 1,384 acres at Fort Bliss and CyrusOne for about 1,201 acres at Dugway Proving Ground in late March 2026.
- Each of the two awarded projects is estimated to cost roughly $2 billion, with developers responsible for financing, construction, operations and eventual decommissioning.
- Initial operating capability is projected for fiscal 2027 at Fort Bliss and fiscal 2029 at Dugway.
- At least a dozen military installations across the Army and Air Force have been opened to private AI data center proposals over the past year and a half, though most remain at the solicitation stage.
- Developers must meet net-zero water usage requirements and provide power plans that do not draw from local electric grids.
Pentagon AI Data Centers
The Defense Department is using underutilized federal land as a bargaining chip to secure access to computing power, especially capacity that can support AI training, inference, logistics and classified military workloads. Under the Army’s Enhanced Use Lease framework, companies would receive access to land while the military gains a strategic digital resource without paying upfront development costs.
The two conditional awards are the clearest sign yet that defense agencies are trying to build a new model for infrastructure procurement. Rather than owning and operating every part of the stack, the Pentagon appears willing to rely on commercial hyperscale operators so long as security, resource use and operational resilience can be addressed. Army officials have framed AI as a force multiplier, making compute capacity a strategic asset rather than a back-office utility.
The market implications extend beyond the awarded sites. Army contracting records point to possible future projects at Fort Hood and Fort Bragg, while the Department of the Air Force has sought proposals for sites in Tennessee, California, New Jersey, Arizona, Georgia and Alaska. That creates potential demand not just for data center operators, but also for private power generation, water management, desalination, cooling technology, secure semiconductors and defense-compliant supply chains.
The Pentagon is effectively trading land for computing power, turning military bases into a new frontier for AI infrastructure investment.
Water, power and political friction
The biggest constraints are not demand for AI compute, but the physical and political limits of building these campuses. Military proposals carry two unusual requirements: net-zero water usage and on-site or otherwise independent power generation that does not tap local grids. Those conditions are designed to reduce pressure on nearby communities, but they also raise engineering complexity and project costs.
Fort Bliss illustrates the challenge. Internal procurement materials reportedly rated water risk in the El Paso area as extremely high. One concept under discussion is for the developer to drill a new well that could support the city’s desalination system, potentially offsetting water consumed by the data center. If such solutions prove workable, defense sites could become test beds for closed-loop cooling, alternative water sourcing and behind-the-meter generation. If not, timelines and returns could come under pressure.
Community acceptance remains another swing factor. In civilian markets, large data center developments have faced pushback over energy use, water draw, noise and land use. Some military parcels under review are remote, but others are near residential or commercial areas. Contract documents show that certain potential sites at Fort Hood and Fort Bragg sit within a mile of civilian housing or businesses, increasing the risk of local opposition and environmental scrutiny.
Implications for Investors
For infrastructure investors, the Pentagon’s push offers a rare blend of federal demand visibility and private financing opportunity. Firms with exposure to hyperscale development, modular power systems, advanced cooling, water recycling and secure digital infrastructure could benefit if more of the currently proposed sites move into awarded status. Alternative asset managers and digital infrastructure specialists may see military land as a differentiated channel for long-duration projects with strategic customers.
At the same time, the policy backdrop is still unsettled. House lawmakers have advanced a provision that would restrict data center equipment containing certain components tied to China, Russia, Iran or North Korea. That could reshape procurement economics for servers, printed circuit boards, chipsets and related systems, particularly if broader sourcing restrictions follow. For listed companies in semiconductors, electrical equipment and supply-chain security, the final legislative language will be an important watch point.
Execution risk also matters. Only two projects have been conditionally awarded, and both still require lease negotiations and environmental review. Investors should monitor whether independent power systems can be permitted and financed at scale, whether water-neutral commitments are technically achievable, and whether local resistance delays construction. The timeline gap between initial operating capability and a fully built campus may also affect revenue recognition and capital deployment schedules.
Still, the strategic logic is hard to ignore. The Defense Department wants resilient compute capacity on secure land, while private developers want large sites and long-term counterparties. If that alignment holds, Pentagon AI data centers could emerge as a meaningful niche within the broader AI infrastructure buildout.
The next phase will hinge on permitting, congressional action and proof that these projects can meet strict water and power requirements. Investors should watch Fort Bliss and Dugway closely, because their outcomes may determine how quickly military-base AI infrastructure scales across the U.S.
VIP Algorithmic Setups
