Deep dives
For readers who already know the basics: how the pieces work, and what to watch.
1. Lattice as an architecture, not a product
Lattice is best understood as three layers: a data layer (entities, tracks and tasks in a shared schema, distributed by Lattice Mesh), an autonomy layer (behaviors that let uncrewed systems plan and coordinate under supervision), and an application layer (C2 interfaces, battle management, third-party apps through the SDK). The strategic question is whether Lattice becomes the default data layer for U.S. and allied forces. The Army's NGC2 common data baseline and NATO's eAirC2 evaluation are the two tests to watch. The developer changelog, which we track automatically, shows the move from a gRPC-first SDK (v1) to an OpenAPI/REST-first SDK (v2) in 2025, with Protobuf definitions published to the Buf Schema Registry: a sign Anduril wants outside developers building on it.
Lattice hub · Lattice vs. Palantir
2. Autonomy and the human in the loop
U.S. policy (DoD Directive 3000.09, updated in 2023) does not ban autonomous weapons. It requires senior review and "appropriate levels of human judgment over the use of force." Most Anduril systems are autonomous in navigation, sensing and tracking, with a human authorizing engagements. The engineering trend is toward supervisory control, with one operator managing many vehicles, which is why the CCA mission-autonomy contract matters as much as the airframe. Watch for how test reports describe operator workload and engagement authority.
3. Layered counter-drone defense
The economics problem: defending against $20,000 drones with $1M+ interceptors fails at scale. Anduril's answer layers cheap detection (Sentry, Wisp, Spyglass, Spark), soft kill (Pulsar jamming) and progressively more capable hard kill (Anvil ramming interceptors, then the reusable Roadrunner-M), with Lattice doing fusion and engagement management. Key programs: the Marine Corps installation counter-UAS program of record ($642M, 10 years), JIATF-401's selection of Lattice for counter-UAS C2, and SOCOM orders. Open questions: performance against large swarms, and how often Roadrunner is actually recovered and reused.
4. CCA: affordable mass in the air
The Collaborative Combat Aircraft program bets that cheaper, uncrewed jets can multiply crewed fighters. Anduril's FQ-44 and General Atomics' FQ-42 won Increment 1 production in June 2026. The Air Force has talked about more than 150 Increment 1 aircraft and a larger fleet over time. The program separates the airframe from the autonomy software, an important structural choice, and Anduril won a mission-autonomy role too. Watch unit cost, production rate at Arsenal-1, and Increment 2 requirements.
5. Munitions and the rocket-motor bottleneck
U.S. munitions production is constrained by a few suppliers of solid rocket motors and energetics. Anduril bought Adranos (2023), received Defense Production Act Title III funding, and pitches itself as the third major U.S. motor supplier. Barracuda is designed around reduced part counts and simpler tooling for high-rate production, with licensed production in Poland. Watch delivered quantities (not capacity claims) and qualification on existing missile programs.
Manufacturing tracker · Barracuda
6. Undersea autonomy
XL-AUVs like Dive-XL / Ghost Shark are designed for long, covert missions: surveillance, seabed warfare, payload delivery. Australia's program of record and a Sydney factory make it Anduril's first major international production program. The U.S. Navy and DIU selected Anduril for an XL-AUV program in 2026. Copperhead (high speed) and Seabed Sentry (persistent seabed sensing) suggest a full undersea product family is forming.
7. The business model under stress
Self-funded R&D works when production contracts follow. Anduril's valuation assumes several programs reach high-rate production at once (CCA, Barracuda, counter-UAS, NGC2) while it builds Arsenal-1. The risk is a timing mismatch between capital spent and revenue received. Company-stated revenue of about $2.2B in 2025 against a $61B valuation implies investors expect years of very fast growth.