US Army's 10-day desert exercise exposes drone and AI warfare gaps
The Wall Street Journal reported that the U.S. Army held a roughly 10-day exercise at Fort Irwin in California's Mojave Desert in July, deploying about 9,700 soldiers in a simulation of combat with Chinese forces to test drones from multiple vendors, AI command systems and next-generation communications gear. Temperatures near 43°C caused servers to overheat, batteries to drain and some equipment to fail to charge, forcing Anduril-built drones to land; some drones needed about 10 minutes to take off after charging, versus under a minute for comparable Ukrainian drones, and flight time was cut to half of maximum endurance. The report added that some drones still relied heavily on manual operation rather than autonomous flight and could not coordinate with traditional equipment — in one simulated attack a five-drone formation turned back because three Black Hawk helicopters entered the same airspace — while the new NGC2 command-and-control system, using satellite communications and real-time data sharing, reportedly cut mission-planning time by about two-thirds.
The exercise suggests that advanced AI models and chips alone do not deliver drone warfare: production capacity, endurance, communications and crewed-uncrewed coordination remain basic engineering bottlenecks.