A Shahed-136 is $20–38k; a Patriot interceptor is >$3M. A $500 FPV can kill an $82.5M F-35. Ukraine’s Spider’s Web: $234k of drones vs ~$7B of Russian bombers. The defender’s intercept economics are broken unless C-UAS is cheap, layered, and recurring.
Oppenheimer (27 Feb 2026) called drones the fastest physical-AI pocket. Grand View: $83.8B (2025) → $182.5B (2033). Warfare subsector $75.8B by 2035. Anti-drone $2.7B → $34B (28.8% CAGR). Lower-skies (C-UAS + tactical UAS + UGS) grows at 2× the broader defence market.
March priced the Iran-war “years of demand into months” story at peak multiples. AVAV $17.7B → $9.6B. DRO $2.5B → $1.3B. UMAC $511M → $1.36B. Thesis intact; entry prices are not March prices.
| Ticker | Layer | Role / what changed since March | Price | Mkt cap | YTD | TTM sales | 1FY sales | TTM P/S | 1FY P/E | Cons. PT | Stance |
|---|
Still the right floor names. Both already did the de-rate the March note warned about (AVAV −46% mcap, KTOS −24%). Street sales still growing (AVAV $2.19B / KTOS $1.78B). Multiples remain rich on EPS (57× / 78×) because earnings have not caught sales. Use as theme ballast, not a new high-beta bid.
ONDS is now a ramp story the street already believes ($530M 1FY vs $51M TTM). 1FY EPS +$0.39 flipping to 2FY −$0.24 is not a clean PE — ignore the 23× and watch bookings vs $530M. DRO is the opposite: profitable, pipeline was A$2.3B, but estimate momentum is negative and the stock has already paid for that. Prefer architecture (detect + defeat + C2) over a single effector.
Original “most attractive risk/reward” call is the one that worked (+51% YTD, street sales $141M / $212M, 4-wk sales +$8M). NDAA 2028 forced-buyer still ahead. Scale and margin path remain the kill. This is the layer Intelligent Mass would call “picks and shovels with physics.”
Regulatory moat (FCC Covered List / T-Motor OFAC) is intact. UMAC already repriced from option to story stock ($1.36B on $11M TTM). RCAT is cheaper on the 1Y tape but still 17× trailing / 10× 1FY sales with negative EPS. Size as if manufacturing can fail.
PL is the strategic ISR anchor (lossy, 24× sales). NXSN is the only profitable, high-margin name in the public drone book (60% EBIT, 40× trailing, street $356M). Cameras ride every airframe. Add both to the live watchlist theme with this wiring.
Not investable. Hivemind-as-Android is the software endgame the public book cannot buy cleanly (PLTR is the expensive proxy). Watch the S-1. An IPO would also re-rate how much scarcity premium AVAV software and PLTR “defence AI” can keep.
Coffee House published into the Iran-strike tape (Operation Epic Fury / Roaring Lion). Five months on:
Wired three places so the theme is not trapped in one app:
Obsidian theme note: US Drone Dominance.md (updated 2026-08-14).
Shield AI $5.3B (Hivemind). Anduril $61B / $2.2B sales (Lattice + mass). Neros $120M+ (Blue UAS Archer). PDW $200M+ incl. $35M from ONDS (Army C100). If the private cohort IPOs, public scarcity proxies (RCAT / UMAC / ONDS) lose the “only way to buy US drones” bid.
FCC / NDAA China-drone rules reversed · ONDS $530M street collapses back toward $50M · AMPX misses NDAA 2028 qualification · UMAC never reaches Chinese motor cost-parity · DRO pipeline does not convert (revisions already warning) · Neros/PDW IPO absorbs the premium · exquisite-only procurement returns.
Bookings vs street sales (ONDS $530M, RCAT $151M, UMAC $56M) · DRO 4-wk revisions · AMPX Fremont / NDAA line · UMAC US motor plant · Phase II Gauntlet winners · Shield AI S-1 · Switchblade reorder cadence · NXSN margin hold at $315M+ guide.