Pentagon AI use jumped 1,775% in just one year

The U.S. Department of War’s use of artificial intelligence has exploded by 1,775 percent over the past year, jumping from roughly 80,000 users to approximately 1.5 million across a workforce of more than 3 million personnel, the department’s top technology official disclosed at a special operations conference in Tampa, Florida.

Emil Michael, undersecretary of war for research and engineering and the War Department’s chief technology officer, made the disclosure during a panel at Special Operations Forces Week 2026, a three-day annual convention that draws senior military, government, and industry figures focused on the future of special operations and defense technology.

The scale of that growth is worth sitting with for a moment. Eighteen months ago, the Pentagon’s AI user base was roughly the size of a mid-sized American city’s workforce. Today it has expanded to a number comparable to the entire population of Philadelphia, all of them inside one of the world’s largest and most complex institutions, using AI tools that range from enterprise productivity software to classified intelligence analysis platforms. Michael acknowledged that translating rapid commercial AI advances into military applications remains genuinely new territory, even for an organization that has been studying the technology for years.

“You can see that every month, we have a new sort of amazing advancement in AI, and how we translate that down to the Department of War and to the warfighter is a fairly new concept,” Michael said.

The department structures its AI use across three distinct layers. The enterprise level covers administrative and organizational functions, the kind of productivity and data management applications that large corporations began deploying years ago. The intelligence level applies AI to the analysis of sensor data, signals, imagery, and other inputs that feed into military decision-making. Michael described the warfighting level as the most important of the three, and his language about it was direct about what the department is trying to achieve.

“[We’re] embedding [AI] into our systems, so that warfighters can use it to be more precise, to be faster, to make better decisions and [to] bring combat power to the battlefield in a way that shortens time of any conflict, protects our warfighters to the maximum extent possible … and, frankly be as lethal as possible — which is what our warfighters want to do,” Michael said.

The drone program Michael cited as a concrete example of AI’s warfighting application carries numbers that underscore how seriously the department is moving. The Pentagon’s Drone Dominance Program has allocated $1.1 billion for the purchase of 200,000 small lethal drones by 2027, a procurement scale that would have been considered science fiction in defense planning circles a decade ago. The program is also being designed to break the monopoly that a small number of approved vendors previously held over drone sales to the military, a dynamic that Michael described with unusual candor. Under the old structure, authorized vendors rarely invested in upgrading their products because they faced little competitive pressure. The department is now deliberately restructuring the acquisition landscape to change that incentive.

“So, we kind of blew all that up, and the idea [now] is to create a scenario where we could use this money, take all the vendors out there [and] give them a shot [at] becoming one of the [DOW’s] prime vendors of drone and counter-drone capabilities,” Michael explained.

That competitive restructuring matters beyond the drone program itself. One of the persistent criticisms of Pentagon technology acquisition has been that entrenched vendors with established security clearances, certified supply chains, and existing relationships effectively locked out smaller, faster-moving companies that might offer better technology. Opening the prime vendor competition to a broader field of participants applies commercial market logic to a procurement ecosystem that has historically resisted it, and the drone domain, where commercial innovation cycles run months rather than years, is precisely the area where that logic has the most immediate military relevance.

The talent question Michael addressed is equally significant and considerably less solved. The War Department cannot match Silicon Valley compensation packages, and the AI engineers and data scientists whose skills the military most needs are among the most sought-after workers in the modern economy. Michael’s answer is to compete on mission rather than salary, framing service to the department as a patriotic opportunity that also builds career credentials transferable back to the private sector.

“What we’re trying to do is create sort of a … more patriotic point of view, or option, for kids coming out of college or out of grad school to come do something for their country — with their country — and then have that be valuable, so that [if] they go back to a private sector later, they don’t feel like they’ve lost time or energy,” Michael explained.

Several hundred recent graduates have already been hired under this approach, and Michael said he hopes to bring in several hundred more before the end of the calendar year. The other panelists, Tara Murphy Dougherty, CEO of a defense acquisition software company, and Peter Tague, managing partner at a growth-stage venture capital firm working with the War Department, both reinforced the same theme from the industry side. Tague noted that working on defense missions has become culturally attractive to young talent in ways that were not true a few years ago.

“It is now cool again to care about your country and to work on things related to this mission. We have had no trouble attracting talent, and frankly, the mission angle has been a cherry on the top,” Tague said.

Dougherty was blunter about how her company communicates its priorities to new employees arriving at orientation.

“We are here to build a business that serves the warfighter. If that is not a mission you are interested in supporting, call somebody else,” she said.

The 1,775 percent growth figure Michael cited is a measure of adoption, not capability, and the gap between those two things matters. Having 1.5 million department personnel using AI tools does not automatically mean those tools are being used effectively, that the outputs are reliable, or that the warfighting applications are anywhere near the maturity of the enterprise productivity uses that likely account for a significant share of that user count. What it does confirm is that the institutional resistance to AI adoption that slowed the Pentagon’s early engagement with the technology has largely dissolved, replaced by an organizational push from the top that is moving faster than the department’s own leadership expected when Michael took the job less than a year ago.