GUEST BLOG: The DoD must emulate Ukraine’s iterative edge

Blog

March 10, 2026

By now, “Ukraine success = UAVs + AI” [artificial intelligence] is not only a meme to American warfighting planners and U.S. Department of Defense (DoD) leadership, it’s the principal catalyst for how the U.S. military wants to speed up program and product development.

In a recent article, “Lessons from Ukraine : Battlefield Drone Innovation Redefines Modern Defense, defense opinion writer Anna Iovenko raises excellent points not about the technology Ukraine is deploying per se, but the process by which that technology is rolled out, tested, upgraded, or tossed out when it proves unworthy. Ukraine has mastered a cycle of rapid rollout, real-time testing, and the immediate  discarding of tech that fails to survive the electronic warfare (EW) environment. U.S. defense planners and designers desperately want to emulate this process, especially the use of AI to offload warfighters and supplement them – if only to deal with reams of sensor data now firehosed from the finally-digitized battlefield.

In fact, the problem of too much sensor data has grown acute, Col. Jeff Pickler, Second Multidomain Task Force and Deputy Commander of the 56^th Multidomain Camp Command Europe, recently shared in a U.S. Army media roundtable on Dynamic Front exercises. Pickler said soldiers are drowning in so much battlefield data that AI is needed to make sense of it all. “There are not enough people that we could stuff in a headquarters or a command post that will ever be able to fully process all of that. It's going to take automation.”

Yet in both of these areas – how to rapidly and cheaply get the best technology onto the battlefield like Ukraine, and where and when to apply AI for meaningful results – the DoD is struggling. In fact, many insiders, prime contractors, and even the services are pushing back while citing the tried-and-true MIL-SPEC model of research and development, development, funding, testing, and deployment. With AI, we see decisions being made more for optics purposes (we’ve got to have AI … something!) rather than a thoughtful approach to what makes sense.  

Truthfully, the bottleneck on the technology side isn't the AI algorithm but the compute density required to run it at the tactical edge. Converting raw sensor streams into actionable intelligence requires massive throughput in a form factor that doesn't melt in a Humvee or fail under high-vibration environments. The U.S. cannot afford to wait for the “perfect” program of record while adversaries iterate in weeks. For battle-ready computation, the U.S. and its allies need to deploy rugged, scalable processing power today and stop treating AI as a buzzword and start treating it as a hardware-integrated reality.

Regardless, the DoD absolutely must take lessons from Ukraine and should consider throwing the baby out with the bathwater. That is: Intentionally break some of the older, staid development and deployment methods while avoiding going whole-hog in on AI “anything” for the sake of marketing-speak.

Chris Ciufo is president and CTO of GMS, Inc. (General Micro Systems).

GMS · https://www.gms4sbc.com/