Battlefield AR could become a command interface, not just a soldier’s display

Anduril’s latest pitch for military augmented reality is no longer just a display in front of one eye. According to MIT Technology Review, the company is working with Meta on prototype AR glasses that would fit into existing military helmets under the $159 million Soldier Borne Mission Command program. The intended shift is blunt: a soldier would not need to pull out a tablet, hunt for the right radio channel or manually reconcile data from several systems. Maps, drone positions, target recognition and command context would move directly into the field of view.

The core of that stack is Anduril’s Lattice software, a platform built to merge sensors, autonomous systems and military hardware into a common interface. That software strategy gained more weight after a reported $20 billion Army integration contract, because Anduril is not simply selling smart glasses. It is selling a model in which battlefield data becomes one operating layer. Meta brings experience in wearable optics and AR hardware, but military conditions are harsher than consumer electronics: battery life, heat, weight, resilience, security and usability under stress are not finishing details. They decide whether the device helps or becomes another liability.

The most provocative part is not the map overlay. It is the command model. Quay Barnett, an Anduril vice president and former Army Special Operations Command veteran, has framed the goal as optimizing the human as a weapons system. In practical terms, a soldier could use voice commands, gaze tracking and subtle taps, while large language models such as Gemini, Llama or Claude process parts of the interaction. That creates a real tactical possibility: faster target marking, drone tasking or sensor filtering without a traditional screen-and-menu interface.

The same mechanism carries the central risk. If the system misunderstands a command, compresses the situation too aggressively or fills the soldier’s view with too many icons, it does not reduce cognitive load. It relocates that load into the helmet. The history is not empty here: Microsoft’s IVAS military headset effort, tied to a $22 billion contract, showed how difficult it is to turn an AR demonstration into equipment soldiers actually want in field conditions. Anduril is now trying to avoid that wall through its own software layer, Meta’s hardware experience and a narrower focus on mission command.

SBMC is not expected to move into production before 2028. Anduril is also self-funding EagleEye, a broader helmet-and-headset concept that the military did not initially request. That matters because the company is not merely waiting for a government requirement; it is trying to define the category. If it works, military AR will not be an extra display. It will become a new mediator between a person, a drone, a sensor network and a decision to use force. If it fails, it will underline the same hard lesson: in battlefield automation, the hardest interface is still the human one.