Defense-tech company Anduril, in collaboration with Meta, has revealed new details about its augmented-reality smart glasses for the US military, designed to attach to existing helmets and enable drone strikes via eye-tracking and voice commands. The project, part of the Army’s Soldier Born Mission Command (SBMC) program, secured a $159 million prototyping contract last year, while Anduril also self-funds a separate helmet-headset combo called EagleEye.
Quay Barnett, Anduril’s vice president leading the effort, described the goal as optimizing 'the human as a weapons system.' The glasses overlay battlefield data—such as maps, drone locations, or AI-identified targets—directly onto a soldier’s field of view. Soldiers can issue commands in plain language, like ordering evacuations or planning routes, with an AI model translating speech into executable software instructions. Anduril is testing Google’s Gemini, Meta’s Llama, and Anthropic’s Claude for this role, despite Anthropic’s public conflicts with the Pentagon.
The SBMC program, awarded to Anduril and Meta in 2025, aims to replace Microsoft’s canceled $22 billion HoloLens contract, which was scrapped after the glasses failed to meet military requirements. The Army does not expect to move its top SBMC prototype into production until 2028, if at all. Meanwhile, Anduril’s self-funded EagleEye project, announced in October 2025, proposes a custom helmet-headset system the company claims the military will eventually prefer, though no formal request has been made.
Anduril’s software platform, Lattice, serves as the backbone for the smart glasses, integrating data from diverse military hardware into a unified interface. In March 2026, the Army announced a $20 billion contract to integrate Lattice across its entire infrastructure. The system is designed to execute multi-step tasks, such as deploying drones for surveillance, identifying targets like artillery units, and recommending tactical actions, like strike coordination or route adjustments.
The smart glasses’ capabilities extend beyond passive data display. Soldiers can use voice commands to direct drones, request real-time intelligence, or even authorize strikes. For example, a soldier might instruct a drone to surveil an area, return upon identifying an artillery unit, and then receive AI-generated recommendations for engagement. The system’s reliance on large language models (LLMs) aims to streamline communication between soldiers and machines, reducing cognitive load in high-pressure environments.
Anduril’s partnership with Meta leverages the latter’s expertise in augmented-reality hardware, though the project remains distinct from Meta’s consumer-focused products. The collaboration reflects a broader trend of tech companies pivoting to defense contracts, particularly in AI and mixed-reality applications. However, the project’s timeline underscores the challenges of military adoption: even if Anduril’s prototypes succeed, production is not guaranteed before 2028.
The cancellation of Microsoft’s HoloLens contract in early 2025 highlighted the difficulties of deploying consumer-grade AR technology in military contexts. Anduril’s approach prioritizes ruggedization and battlefield-specific features, such as real-time threat assessment and seamless drone integration. Barnett emphasized that the goal is not just to provide information but to create a 'shared consciousness' between soldiers and autonomous systems, enhancing decision-making speed and accuracy.
Despite the ambitious vision, the project faces skepticism. The Army’s history of failed AR initiatives, including the HoloLens debacle, raises questions about whether Anduril’s prototypes will deliver on their promises. Additionally, the reliance on commercial AI models like Gemini and Llama introduces potential vulnerabilities, such as latency or misinterpretation of commands, which could have life-or-death consequences in combat scenarios.
ops.llm_calls. Every fact traces to a citation. If a fact looks wrong, write to corrections.