On June 16, Empower AI announced it had retained a $255.4 million, five-year recompete contract from the Defense Information Systems Agency to automate enterprise IT support for the Pentagon and the National Capital Region. The scale is worth pausing on: DISA J6 operates the IT infrastructure that connects the office of the Secretary of War and all Pentagon Fourth Estate agencies, the network backbone carrying classified and unclassified traffic for 300,000 users. This is not a pilot. It is not a proof-of-concept with limited scope or a two-year test. It is a vote of confidence in a commercial company's ability to use AI to reduce labor, accelerate problem resolution, and lower total operational cost across one of the most cyber-sensitive IT environments in the federal government.
The contract structure tells you how serious DISA is about the bet. A one-year base period followed by four, one-year options means DISA can evaluate Empower AI's actual performance against promised automation metrics before locking in the full five-year value. CEO Jeffrey Bohling's statement, 'We will apply AI-enabled automation where it measurably improves outcomes for DISA J6 mission partners, while maintaining the reliability and security that DISA J6 Service Delivery operations require', is not boilerplate. It is a hedge, and a necessary one. DISA is not betting that AI will solve everything; it is betting that targeted automation in specific, measurable use cases will work. The contract emphasizes 'reducing manual work' and 'speeding up resolution times for users,' which are concrete operational metrics, not aspirational claims about transformation.
This recompete matters because it is a recompete. Empower AI already held this contract. The fact that DISA renewed it, and by all available accounts, expanded the automation scope, is the signal. Federal IT procurement is not fast. DISA had the option to rebid the work. It chose to retain the incumbent and scale the AI automation mandate. That is adoption, not experimentation. The contract also signals a shift in how DISA evaluates vendors. In the previous cycle, Empower AI likely won on service delivery capability and cost. In this one, AI-driven automation is now part of the evaluation criteria itself. Any competitor bidding DISA J6 work going forward will be expected to articulate how they use AI to reduce operational overhead.
The competitive landscape matters here too. This is not a TAI or Palantir contract in the classic sense, DISA is not buying custom AI models or specialized algorithmic capability for intelligence analysis. Empower AI is buying the right to deploy enterprise IT automation across federal IT operations. That is a smaller, less glamorous market than defense analytics or weapons targeting, but it is deeper than it looks. Every federal agency with large IT footprints, VA, HHS, Interior, Treasury, DHS, faces the same problem: labor-intensive ticket resolution, slow deployment cycles, high operational overhead. If Empower AI can demonstrate measurable labor cost reduction and uptime improvement at DISA over the next 18 months, it has a replicable template for nine-figure contracts across the federal government.
Two specific metrics will determine whether this bet actually pays off. First: whether Empower AI achieves measurable reduction in mean-time-to-resolution (MTTR) for IT support tickets across DISA's network by end of FY2027. DISA has 300,000 users; if automation cuts average ticket resolution time by 15 to 20 percent, that is a justifiable contract renewal and a proof point for replication. Second: whether the automation footprint actually expands in year two, or whether DISA pulls back to defensive automation only after encountering edge cases or reliability issues. The base period is the test. If automation scope shrinks in option year one, the bet did not work. If it expands, the model holds.