Anduril and Meta are building smart glasses for soldiers, and according to MIT Tech Review, the system would mark a major escalation in how closely frontline troops rely on imperfect AI. The plan: helmet-mounted displays that identify threats, recommend strikes, and run powerful computer vision locally when 5G isn’t around. This is the first time most infantry would carry AI that whispers targeting suggestions in real time.
The backstory has a twist. In 2017, Facebook pushed out Anduril founder Palmer Luckey over an internal fight tied to his support for Donald Trump. Eight years later, MIT Tech Review reports, the two companies are partners again, with Meta supplying displays and waveguides while Zuckerberg warms up to the second Trump administration. Politics and procurement don’t usually rhyme this cleanly.
A crowded race, not a solo run
Anduril isn’t alone. Rivet just landed a $195 million prototyping contract. Israeli defense-tech firm Elbit picked up $120 million in March. All of this happened after Microsoft lost the Army’s smart glasses program following a Pentagon audit that found testing gaps serious enough to potentially waste $22 billion.
That’s the real signal here. The Army is willing to burn vendors and restart contracts to get this right. Three serious players now circle the same pie, which means pricing pressure, faster iteration, and probably one ugly public failure before someone wins at scale.
What’s actually new in the tech
Anduril is testing digital night vision for both of its prototypes. Electronic sensors plus algorithms boost low light, replacing the analog tubes soldiers have used for decades. The promise has been around forever. The execution has not. Anduril says it’s mixing generative AI with older machine learning to finally make the images sharp and fast enough for combat.
Two form factors are in play. One bolts onto existing helmets and gear with a separate battery pack. The other, called EagleEye, builds the tech directly into the helmet. If the Army passes on EagleEye, Anduril plans to shop it to foreign militaries anyway. That’s a hedge worth noting. Defense startups are no longer betting everything on a single Pentagon contract.
Why this matters beyond defense
Computer vision in militaries isn’t new. Chatbots even crept into decision-making during the recent war in Iran. What’s new is putting AI suggestions one centimeter from a soldier’s eye, in dust, smoke, and explosions, with no cloud connection. Every assumption that consumer AR companies make about clean environments and reliable bandwidth gets stress-tested here.
The engineering bar is brutal:
- Operate through dust, blast pressure, and smoke
- Add capability without piling onto the 100+ pounds soldiers already carry
- Run heavy AI models locally without 5G
- Stay seamless enough that troops trust the overlay under fire
If Anduril and Meta crack any of these, the spillover into consumer AR, industrial wearables, and emergency response is huge. Edge AI that works in chaos is the same problem firefighters, surgeons, and warehouse operators have been waiting on.
What to watch over the next 18 months
- The first field test failure. Someone’s prototype will misidentify a target. How the contracting agency responds will set the rules for AI accountability in combat.
- Meta’s dual identity. A company selling Ray-Bans to teens and waveguides to infantry will face brand questions. Watch how Zuckerberg threads that.
- Foreign sales. Anduril already signaled it’ll sell to allies. That accelerates a global market for battlefield AR with very little regulatory scaffolding.
- Edge AI breakthroughs leaking out. Whatever makes local computer vision work in a war zone will be sitting in a Best Buy product within five years.
The Pentagon analyst quoted by MIT Tech Review put it plainly: “It’s got to work, and it’s got to be pretty seamless. It’s a high bar.” That bar will define how soon AI rides on every soldier’s face, and which company gets to set the standard for the rest of us. Full details at MIT Tech Review.