An artist's composite in emerald and teal light, layering a proposed uncrewed aircraft, a production campus, an orbital sensor platform and a hypersonic track into one night frame.

Proposals

Six Programs, Already On The Table

Each one is a complete, costed engineering proposal against a problem the force is carrying right now. None has been built. Every figure below is carried straight from the proposal that produced it, with the maturity of that figure stated beside it.

A vast munitions manufacturing campus at night from the air: four parallel production halls glowing from within, a minerals processing plant, and a worked extraction terrace stepping down into the hillside.

AI Managed Missile Factory

Dirt in one end, rounds out the other

Four production lines on one reconfigurable campus. Sized for 11,047 rounds a year — the second source that extending a prime’s line cannot give you. Designs stay government-owned the whole way through. And the campus brings its own rare earths into production in roughly half the Western precedent.

Rounds a Year 11,047 MODELLED all four lines at OEE 0.55 — a sizing basis, not a production commitment

PAC-3 Rounds a Year 2,462 MODELLED up from a record 620 in 2025

Rounds for the Money 4 to 1 MODELLED $25.2B against $5.6B of operating cost on the 6,322 rounds priced; Javelin has no sourced price

Rare Earths on Site 8.35–11 yr MODELLED best case to 50th percentile, against 16.9–17.8 yr Western precedent

$304M P50 · Phase 0 only · go/no-go, government-owned Read more →

A large space-based optical sensor platform in orbit, solar wings extended, watching a hypersonic vehicle running through the upper atmosphere far below on a long incandescent plasma trail.

AEGIS-ARGUS-KRONOS

Closing the gap radar cannot cross

The instinct has been to chase the threat with more radar. This one looks instead. Space-based optical sensing above, layered effectors below, and an AI-fused shoot-look-shoot chain joining the two.

Discrimination First Cut The Count SPECIFIED rejecting decoys buys more than chasing a high single-shot number

Sensing Approach Optical, From Orbit SPECIFIED the plasma sheath that defeats radar is the infrared source this is built to see

Sensor Custody Two Sensors Or No Shot SPECIFIED fire-control-eligible only while two independent sensors hold line-of-sight

Discriminants Four Votes SPECIFIED ballistic-coefficient, thermal-inertia, micro-Doppler and optical spectral, fused

First article · constellation excluded · interceptor rounds buyer-furnished Read more →

An uncrewed aircraft with a long slender wing banking above a cloud deck at golden hour, carrying several small slim interceptor bodies on underwing rails, with a second aircraft holding station far below.

IRON HIVE

A reusable flying magazine above the threat

On the night of 13–14 April 2024, one combined raid of roughly three hundred weapons drew high-end interceptor magazines down across an entire coalition in a single night. Each round spent cost orders of magnitude more than the thing it stopped. IRON HIVE is designed so that trade stops being ruinous.

Single-Shot Kill 69% MODELLED at 1.3 m guidance error against a ~2 m lethal radius; 96% at 0.8 m, 34% at 2.2 m

Reach Lofted 15–30 km MODELLED for the slender high-ballistic-coefficient body near 8,400 kg/m²

Cost Per Kill $19–26k MODELLED all-in at 1.4 rounds a kill with carrier and radar amortised; an AIM-9X-class round is near $400k

Endless Coverage Recovered & Reused SPECIFIED netted, flown back, inspected and flown again

Phase 1 · G/NG-1 cost gate · prototype instrument Read more →

A line of slender sensor masts rising from coastal water at dusk, with a small four-rotor interceptor climbing hard in the foreground toward a distant hostile drone, and the softened grey side of a warship in the haze behind.

DRONE WALL

Two pieces designed to work together

This one does not set out to replace the ship’s air-defence system. It asks a narrower question: whether a cheaper layer can preserve high-value magazine depth against a bounded slice of the threat. Nothing has flown, no blocker has captured anything, and no host vessel has authorised it.

Threat Speed Bound 80 m/s MODELLED an analysis boundary set above the reference threats — not a permanent one

Detection Declaration 2.0 s SPECIFIED proposed acceptance criterion; no measured basis yet, confidence stated LOW

Abort Response 1.0 s SPECIFIED proposed safety-command allocation; no approved threshold exists yet

Development Labour $22.88M ESTIMATED burdened direct labour, the same in both profiles because development is non-recurring

$85,663,051 · Profile A · prototype gates, not fielding · cost confidence LOW Read more →

A sealed matte grey line-replaceable unit with milled cooling fins mounted on the wing pylon of a large uncrewed aircraft, photographed from below at night on a wet hardstand under emerald floodlight.

FANGS

Assured PNT and EW in one airborne unit

Everything the aircraft needs to know where it is and keep talking, inside one sealed box. Modelled end to end for GPS-denied airspace and a contested spectrum.

GPS-Denied Leg 100 m CEP MODELLED aided bound over ten minutes; unaided inertial coast degrades to 2.33 km

Contested-Link Design Four Ways Back SPECIFIED automatic link establishment, space diversity and mandatory relay — degradation, not loss

EW Transmit Span 0.5 to 6 GHz SPECIFIED Vivaldi across 1–6 GHz with an LPDA covering 0.5–2 GHz, continuous

Internal AI Pilot Plans the Mission SPECIFIED state what the aircraft is to achieve and it works out how to carry that out

Funded PDR · Tasks 1, 2 and 4 · Task 3 plan for approval Read more →

A long final-assembly bay in an older defence plant at night shift, CNC machining centres and robotic arms in motion, with a faint emerald luminance tracing the contours of the existing machinery and the path between stations.

VES · Virtual Execution System

Multiplying output on the factories we already have

This one does not replace anything. A retrofit layer slides underneath the processing already running on the floor, learns it, and optimises it without touching the critical path — so an incumbent MES becomes worth more instead of being torn out. Existing control stacks. Unchanged article designs. Eighteen months.

Output Multiplier 1.6–3x MODELLED output at 85% equipment availability, on unchanged designs

Unit-Cost Saving 22% MODELLED against the Month 18 baseline, Javelin final assembly

Cycle Throughput 2.0x MODELLED per-article cycle time 48 minutes to 23 at Month 18

Schedule Advance 9 months MODELLED on the MRL 8 low-rate-initial-production gate, Month 18 modelled

$424.9M · 60 months · three factories · first tranche $47.9M over 18 months Read more →

Modelled — produced by analysis under stated assumptions Specified — a design requirement the system is built to meet Estimated — a costed figure carrying commercial uncertainty Objective — a target the programme is aiming at

Next step

Ask for the one you want to read

Every programme above has a full proposal behind it — the modelling, the cost basis, the gates and the assumptions in full. Tell us which one and we will send it.

Request a proposal →