Thematic long · Defence · Autonomy + Intelligent Mass + Capacity

The New Arsenal — Intelligent, Networked Mass

Framework adapted from Anduril CEO Brian Schimpf’s investor letter (released with the Series H). The market already knows the world is rearming. The better question is where value accrues as militaries change what they buy and how they fight — finished platforms, the software connecting them, munitions capacity, or the scarce components that make mass possible.

Source: Anduril Series H · May 2026 $5B raise @ $61B valuation 2025 rev $2.2B (~2×) Lead: Thrive + a16z Public expressions: LHX · RTX · KTOS · AVAV · PLTR

Market data: Bloomberg ~2026-08-09 · Theme synthesis: Inflection Economy (2026-08-07) + Anduril letter · Not investment advice

Anduril valuation

$61B

~28× 2025 sales (software-like)

Anduril 2025 revenue

$2.2B

More than doubled YoY

LHX Missile Solutions

$1B

DoW convertible preferred; IPO H2’26?

Link to DIB book

Munitions chokepoint names on sibling tab

Inflection 1 — AI + autonomy

Advantage shifts from a few exquisite platforms toward distributed, software-defined systems. Fewer people control more platforms; cheap long-range strike holds high-value assets at risk deep behind the front.

Inflection 2 — Industrial reality

Great-power competition exposed that the West cannot replenish munitions and equipment at wartime speed. Production capacity becomes part of the weapon system — factories, qualified suppliers, test infrastructure, software update cadence.

Inflection 3 — Procurement + capital

Reform and industrial policy direct capital toward scalable production, scarce components, and the software layer that connects the battlefield. Startups can win programs of record; primes still set demand and bear much of the capex.

Schimpf’s label: “intelligent, networked mass.” High-end platforms do not disappear (subs, stealth aircraft, satellites, missile defence). They operate alongside large numbers of cheaper sensors, drones, decoys, EW systems and effectors that must be manufactured quickly, coordinated through software, and replaced when lost. The moat may not be the weapon itself — it may be the ability to build the next thousand.

Where value may accrue — the stack

Read bottom-up for bottlenecks (hardest to expand); top layers get the headlines and often the richest valuations.

Platforms

Autonomous air / UAS / CCA

Visible · competitive

Lower-cost unmanned combat aircraft, attritable drones, loitering munitions. Strategy closest to Anduril’s “mass” thesis in public markets: KTOS (Valkyrie / jet drones), AVAV (drones + BlueHalo stack). High thematic purity, often valuation- and FCF-dependent.

KTOS · AVAV
Anduril (private)

Software

C2 / data / Lattice layer

Still up for grabs

Operating layer that fuses sensors, platforms and effectors under degraded comms. Palantir = neutral integrator. Anduril Lattice = hardware-linked vertical. AVAV mission software + prime proprietary C2 + government open architectures all compete. Category is valuable; structure is unsettled.

PLTR · Lattice (priv)
AVAV · primes C2

Infrastructure

Counter-UAS

Recurring defence infra

Cheap attack drones create an economic problem for defenders. Solution is layered: RF/radar detect, EW, directed energy, kinetic intercept, software choosing the cheapest kill. Looks less like a niche weapon and more like base / ship / site infrastructure.

AVAV · RTX · LMT
specialists

Capacity

Missiles & interceptors

Backlog + factory cycles

Tomahawk / AMRAAM / Standard Missile / PAC-3 / THAAD type production ramps. RTX and LMT carry multi-year industrial programs. Less “pure defence-tech,” often better risk-adjusted than a single-program autonomy name.

RTX · LMT · NOC
GD

Bottleneck

Propulsion · energetics · SRM

Hardest to replicate

Solid rocket motors, oxidisers, seekers, qualified electronics. Platforms can be prototyped; motors and energetics cannot be wished into existence. LHX Missile Solutions (Aerojet Rocketdyne) + $1B DoW convertible preferred and planned MSL IPO is the cleanest public “propulsion capacity” expression. See sibling Chokepoint Dashboard for NEU/AMPAC, NOC SRM, JP titanium, etc.

LHX · NOC · NEU
012450 KS · 7013 JP

Picks & shovels

Sensors · edge · RF · GaN

Often better economics

Every autonomous node needs sensing, secure comms, actuation, power. Fragmented but qualification-locked — same “receives the capex” logic as the munitions book. Overlap with DIB guidance / components layers.

LHX · TDY · MRCY
MTSI · QRVO · …

Six investment implications

From the Schimpf letter via Inflection Economy synthesis — use as a diligence checklist, not a buy list.

1Bottlenecks may beat platforms

More competition at the finished-system layer; concentration at propulsion, seekers, energetics, qualified electronics. Ask: who must every new autonomous design buy from?

2Theme can be right and still overvalued

Anduril proves the category. Intelligent mass still needs physical mass — factories, inventory, fixed-price risk. Public names can trade like software while cash flow looks industrial.

3Software layer unsettled

Lattice vs Palantir vs prime C2 vs open architectures. Neutrality (PLTR) vs vertical integration (Anduril) is the structural fight — not “will software matter.”

4Counter-drone → infrastructure

If sites need continuous protection, budgets look more like recurring defence infrastructure than one-off weapons buys. Architecture winners > single interceptor vendors.

5Undersea autonomy

Deep-sea / undersea mass is a stated next domain in Schimpf’s framing. Public pure-plays are scarce — underwrite carefully; do not force-fit ticker lists.

6Sensors & edge as better risk/reward

Often lower story risk than CCA headlines, higher qualification moats than generic electronics. Cross-read with munitions chokepoint scores on the main DIB tab.

Public expressions (snapshot)

Bloomberg fields ~2026-08-09. Role / stance are analyst framing for this theme, not formal ratings. YTD is price return.

Ticker	Layer	Role in thesis	Price	Mkt cap	YTD	Fwd P/E	EV/EBITDA	FCF yld	Stance

Valuation tension (defence-tech)

KTOS and AVAV align tightly with “affordable autonomous mass,” but trade at high forward multiples with negative FCF yields in this snapshot — exactly the software-multiple-on-industrial-cash-flow risk the letter implies for Anduril itself (~28× sales).

KTOS: ~65× fwd PE · ~48× EV/EBITDA · FCF yld ≈ −1.2% · YTD ≈ −20%
AVAV: ~57× fwd PE · ~30× EV/EBITDA · FCF yld ≈ −1.6% · YTD ≈ −23%
PLTR: ~88× fwd PE · ~65× EV/EBITDA · FCF yld ≈ +0.8% · software economics better, still rich

Capacity / bottleneck side (cleaner cash)

LHX and RTX express motors, interceptors, radar and production lines. Lower story purity, higher cash conversion in this tape — and 2026 YTD price action rewarded them over pure autonomy names.

LHX: ~23× fwd PE · ~14× EV/EBITDA · FCF yld ≈ +5.2% · YTD ≈ −2%
RTX: ~30× fwd PE · ~21× EV/EBITDA · FCF yld ≈ +3.8% · YTD ≈ +22%
LMT: still the exquisite-platform + interceptor franchise · YTD ≈ +22%

How this tab relates to the munitions chokepoint book

Same industrial war, two cuts. The Chokepoint Dashboard answers: where does the Western munitions chain physically narrow? This tab answers: how does AI + autonomy rewrite what gets bought, and which public layers capture that rewrite? The sibling Drone Stack tab is the company-by-company cut of that rewrite (C-UAS, platforms, batteries, NDAA components, ISR) from the Coffee House March 2026 note. Intelligent mass increases demand for motors, energetics, seekers and interceptors — it does not make the chokepoint map obsolete. Use both: theme for “why demand grows and how systems change”; chokepoints for “who must get paid when volume actually scales.”

Anduril (private reference)

Series H $5B @ $61B (May 2026). Platforms + sensors + defensive systems + factories; Lattice as the connective tissue. Best used as a framework company, not a public comp set member. IPO chatter exists; treat valuation as venture-priced optionality on industrial execution.

Kill / invalidate conditions

Procurement reverts to multi-decade exquisite-only cycles; autonomy programs stay demo-scale; fixed-price losses cascade through defence-tech; MSL IPO / capacity expansions slip while multiples stay software-rich; software layer commoditises under forced open standards with no pricing power.

Working checklist

Program of record vs prototype · unit economics at scale · capex before revenue · contract type (fixed vs cost-plus) · FCF inflection date · qualification lead times for any “bottleneck” claim · whether counter-UAS is architecture or project revenue.

Provenance. Anduril funding and revenue figures from company / press (Series H announcement, May 2026). Public-market multiples and YTD from Bloomberg (~2026-08-09). Theme structure synthesised from Schimpf’s letter and secondary research write-ups (incl. Inflection Economy, 2026-08-07). Layer roles and stances are analyst framing for research navigation, not recommendations. For scored munitions chokepoints and 68-name universe see sibling tabs and value-chain.md. Not investment advice.