Army’s AI command system stumbles in desert test—90+ techs evaluated

The Pentagon’s push to embed AI at the tactical edge just got a dose of reality. In a high-stress desert test, the Next-Generation Command and Control system showed how machine-speed sensor fusion can revolutionize battlefield awareness—but the same trial proved that without resilient connectivity and rugged hardware, even the smartest AI is just a paperweight.

The US Army's ambitious vision for an AI-driven, multi-domain battlefield underwent a rigorous real-world test during Project Convergence Capstone 6 in the Mojave Desert last month, revealing both transformative potential and stubborn technical hurdles. Over nine scorching days at Fort Irwin, thousands of soldiers trialed more than 90 technologies, with the Next-Generation Command and Control (NGC2) system at the centerpiece. Designed to fuse data from sensors, drones, and troops across land, sea, air, space, and cyberspace, NGC2 aims to give commanders an instantaneous, shared picture of the battlespace, enabling them to identify threats and coordinate strikes while constantly on the move—breaking free from traditional, vulnerable fixed headquarters.

“

Over nine scorching days at Fort Irwin, thousands of soldiers trialed more than 90 technologies, with the Next-Generation Command and Control (NGC2) system at the centerpiece.

The exercise, held in late July 2026, marked the latest and largest iteration of the Army's Project Convergence experimentation series, which has become a cornerstone of the Pentagon's push to modernize for potential conflicts against near-peer adversaries. Unlike tightly scripted demonstrations, Capstone 6 placed the technology in the hands of soldiers under realistic, demanding conditions—temperatures in the High Mojave pushed equipment to its limits. Leaders hailed the "common visualization" capability as a clear success, noting that the system provided an unprecedented ability to track both friendly and enemy positions in real time. However, the test was far from flawless. According to Axios, soldiers struggled with intermittent connectivity and overheating gear, underscoring the reality that AI and networked warfare are only as strong as their weakest physical link.

The NGC2 system leverages artificial intelligence to process the vast streams of data from a diverse array of platforms: autonomous robots operating alongside dismounted infantry, specialized "launched effects" drones that can locate targets for long-range artillery, advanced ground sensors, and contested logistics networks. This information is federated into a single interface, theoretically allowing a battalion commander to see that an artillery unit two valleys away is best positioned to neutralize a threat spotted by a drone several kilometers ahead. The result is a dramatically accelerated decision cycle—from minutes or hours to seconds—while the force itself stays dispersed and harder for enemy fires to target.

Still, the connectivity failures reported during the exercise highlight a fundamental challenge. The vision of a seamlessly networked military depends on robust, resilient communications in environments where adversaries are actively jamming signals, destroying nodes, and exploiting electromagnetic vulnerabilities. If a soldier in a remote canyon loses his data link because a relay drone overheated or a satellite channel was disrupted, the AI's battlefield calculus becomes incomplete. The overheating problem also raises questions about hardware suitability for desert, jungle, or arctic operations—environments where future land wars may well be fought. These are not simply teething issues; they go to the heart of whether AI-enabled command and control can be trusted in a large-scale conflict.