Defense AI absorbed 44% of all venture capital deployed in a single week. Anduril $5B, Helsing $1.2B. The rest of startup funding shrank by half that same week.
Defense has become AI's new home turf. And Anduril is at the center of it all.
Why $61B, out of nowhere?
Anduril was founded in 2017 by Palmer Luckey — the guy who built the Oculus Rift and sold it to Facebook for $2B. He left Facebook and dove straight into defense with Silicon Valley speed.
The name itself is telling. "Andúril" is the reforged sword from Tolkien's Lord of the Rings. "Rebuild what was broken" — it's a declaration to redesign 70 years of legacy defense systems from scratch.
But Anduril didn't reach $61B by making better weapons. The real reason is Lattice OS — an AI platform connecting every sensor, drone, weapon, and vehicle on the battlefield under one operating system.
Here's the analogy: legacy defense is a collection of Excel spreadsheets — each weapon system isolated, no shared data. Anduril built a cloud platform on top of all of it. Whether it's a drone, a submarine, or a fighter jet — once connected to Lattice OS, they all see each other in real time and coordinate autonomously.
2025 proved the thesis. Revenue doubled year-over-year to $2.2B. The US Army consolidated 120 separate procurement contracts into a single Lattice platform vehicle worth $20B over 10 years. Not because the weapons are better — because the platform connects everything.
The timing clicked too. Ukraine showed that low-cost drones flip battlefield dynamics. The tactical value of AI autonomous systems was proven in actual combat. Defense venture investment hit $49.1B in 2025 alone — up 80% year-over-year.
What's actually different from old-school defense?
The difference between legacy defense and Anduril is more fundamental than it looks. Their business models are completely different.
| Legacy Defense (Lockheed, Boeing, etc.) | The Anduril Way | |
|---|---|---|
| Product development | Wait for government RFP, then build | Self-fund first → then win the contract |
| Core asset | Physical weapons hardware | AI software platform (Lattice OS) |
| Dev speed | 10–20 year procurement cycles | Silicon Valley speed (quarterly updates) |
| Revenue model | Contract payments regardless of outcome | Product performance = more contracts |
| Ecosystem | Closed, proprietary specs | Open SDK (REST/gRPC APIs for third-party integration) |
"Platform beats product" — and it works in defense too. Legacy defense built one great F-35. Anduril built Lattice OS so every hardware addition increases the whole network's value. Network effects, applied to war for the first time.
Compare the competitors and the differentiation becomes clear:
| Company | Core tech | Valuation |
|---|---|---|
| Anduril | Lattice OS (AI platform) + full hardware stack | $61B |
| Helsing | Defense AI software (HX-2 drone) | $18B |
| Shield AI | Hivemind autonomous flight software | $12.7B |
| Lockheed Martin | F-35 and traditional hardware (70+ years old) | $110B (market cap) |
The Anduril Playbook
The Anduril strategy isn't just a defense story. Medical devices, logistics, manufacturing, finance — any industry with fragmented hardware and entrenched legacy players is fair game.
- Design the platform first, not the product
Anduril designed Lattice OS before anything else. Not individual drones or missiles — the AI layer that connects everything. If you design for "the platform that becomes the industry standard," lock-in follows naturally. - Build before you're asked
Legacy defense waits for what the government wants, then builds it. Anduril built first, showed the results, and won the contracts. Taking on your own R&D risk in exchange for "we already built this" is the Silicon Valley move. - Open SDK = ecosystem
Lattice uses REST/gRPC APIs so third parties can integrate their own sensors, platforms, and apps. Every partner you pull into your platform increases the platform's value. Closed systems die fast. - Acquire hardware to fill platform gaps
Since 2021, Anduril has acquired 8 hardware companies — drones, subs, rockets. They built the platform first, then strategically filled in the hardware the platform needed. - Design manufacturing like software
Arsenal-1 can switch between drones, missiles, and submarines on the same production floor based on demand. They designed manufacturing itself like a platform — flexible, reconfigurable, demand-driven.
The Korea angle
The Korean government has set a target of nurturing 100 defense startups by 2030. Hanwha participated in Shield AI's latest round. The race to build "Korea's Anduril" has already started.
