The US Army selects Carlyle and KKR to build two data centers on military bases, set to cost $2B each, as its token usage rises 8x during the US' war in Iran
Army secretary says Iran war underscores need to adapt to AI's growing role in modern warfare — The US Army has selected Carlyle …
Context & Ripple Effects
The Army’s move extends a procurement arc from AI applications into the underlying compute environment. It follows a military push to use AI tools for rapidly analyzing operational data and earlier Army agreements for software and mission systems, including a long-term Anduril procurement framework.
The reported eightfold rise in token usage makes dedicated on-base capacity consequential: the Army is treating AI demand as infrastructure planning, not only as a software-buying decision. Bringing Carlyle and KKR into the buildout also ties defense AI capacity to large-scale infrastructure finance.
First-order effects
- Carlyle and KKR will build two Army-base data centers under contracts valued at $2 billion each, creating dedicated capacity for the Army’s growing AI workloads.
- The Army gains an on-base infrastructure path as operational AI use expands, while its AI vendors can orient deployments around a more durable military compute footprint.
Second-order effects
- Defense AI suppliers face greater incentive to design offerings around Army-controlled data and compute environments rather than relying solely on distributed or external capacity; this builds on the Army’s device-level AI deployment for threat identification.
- The contracts make private-capital participation more central to military AI infrastructure, potentially widening the competitive field beyond traditional defense primes for long-duration facilities projects.
Third-order effects
- If usage growth persists, defense AI procurement may increasingly bundle models, software, hardware and secure compute into integrated, multi-year capability programs rather than discrete software contracts.
- The pattern points toward strategic compute becoming a procurement category in its own right, with control over infrastructure and financing shaping which AI systems can be operationalized at scale.
The trend: Military AI adoption is shifting from buying individual tools toward financing and securing the dedicated compute infrastructure needed to run them in operations.