99% of Drone Companies Will Die & Why Anduril’s Products Aren’t an Ethics Debate | Matthew Steckman

Summary

• The headline $20B US military contract is a credit-card limit, not revenue: Steckman defines it as “a credit card limit” — “there’s no obligated money,” but the government has created a process in which financing, contracting, and evaluation happen once so every office can buy Anduril’s commercial tech without repeating the process. Dollars only flow on delivery (“delivery is revenue to us”), and the scale context matters: Anduril will do “a couple billion in revenue this year” against Lockheed’s ~$100B — still small, but the vehicle signals the government “thinks that we can deliver some pretty major stuff now and in the future.”
• Drone investing is monopoly-or-die: in every defense technology class “there’s probably one, sometimes two actual programs that if you capture them, you have a business and if you don’t, you have no business.” Of the 30-40 European drone companies Harry has seen, “there’ll be one that wins… the challenge for investors is actually figuring out which one it is.” The most common red flags: hubris (“in the 1960s we invented that”) and overestimated TAM — a single large contract “is not an enduring business.”
• “You basically can’t have a defense company if you don’t have a large US business” — 50% of global defense spend is American, and Europe isn’t a union in defense: “if you’re a French company, Europe kind of isn’t your market — France is.” Any European next-gen prime thesis that doesn’t include US sales is cutting off half the market from the get-go.
• The missiles bet is the surprise winner, validated by the Iran war: Anduril’s Barracuda cruise missiles are built “like you build a bathtub,” letting ordinary US contract manufacturers handle production — “you can handle elasticity of demand for the first time” in a missiles business where a large percentage of offensive and defensive missile systems have been “expended” and are “really hard to replace.” Company-wide gross margins run 40%+ — “in hardware and defense quite good” — with high-volume products lower-margin because of customer expectations for what they should cost over time and the ability to regain margin through cost savings.
• IPO is “in the window”: Anduril wants to go public because “when you are a public company, there is an additional level of trust that is afforded to you” by the national security apparatus. The gate: only ~a quarter of its 20 P&Ls are in rate production throwing money back to the business; the rest are in J-curves Anduril compresses from the industry’s 7-10 years to 3-5 (Roadrunner: 24 months from napkin to fielded, >$100M before revenue). Steckman pegs Anduril’s multiple at “10 or 14x forward” vs peers at 20-40x — Harry’s correction: “I see like 200.”
• VC valuations have put defense M&A out of reach: Anduril has bought “almost a dozen companies” but only traditionally-priced ones — “there are lots of VC companies I would love to buy… but the multiples today put it out of reach.” The category he’d run at with an unlimited checkbook: space domain, where a gap yawns between slow legacy primes and SpaceX’s commercially-adjacent dominance. Trey’s warning stands: “never be surprised around how long a company can continue to exist.”
• On ethics, Steckman says the answer is clear: “we have a democratically elected set of governments that represent the US, the alliance, NATO… they set the rules and we abide by those rules” — second-guessing that framework creates “a pretty dramatic set of slippery slopes.” And war isn’t good for business over the right horizon: companies “myopically focused” on that conflict “will not have a business coming out the other end.”

Deep dive

1. The $20B contract, decoded: a friction-killing credit-card limit

• Steckman’s own definition, against “a rainbow of interpretations” online: “think about this as a credit card limit. The government has said up to $20 billion they can spend on a class of Anduril’s commercial technology. There’s no obligated money.” What it removes is repetition — financing, contracting, and evaluation that each individual office would ordinarily run “now only have to happen once. And they just did.”
• On what “dependent on delivery” means: “delivery is revenue to us — when I deliver something I recognize the revenue; same thing in government, when we deliver something the actual procurement orders and dollars flow.” He resists “fast track”: “it’s more like in the arduous process to actually get something contracted, we’ve cut a lot of the first steps out.”
• The signal value exceeds the number: the vehicle means “we have enough business with the federal government to justify putting together a fairly sophisticated contracting vehicle… they are not going to do that unless they think it’s worth it.” Scale check — Lockheed does ~$100B a year in revenue; Anduril “a couple billion this year.”

2. You can’t be a defense company without America

• The categorical claim: “you basically can’t have a defense company if you don’t have a large US business — 50% of defense spending is in the US and 50% is in the rest of the world. If you are cutting off half of your market from the get-go, that’s probably a big problem.”
• The Europe trap, in response to Harry’s jab that “the French military spend less than I do in Chanel for my mother”: Europe “is not a European Union in defense spend — each country has its own sovereign companies and its own sovereign agenda. You keep getting winnowed and winnowed… if you’re a French company, Europe kind of isn’t your market. France is.”
• The two red flags Steckman sees most in emerging defense startups: hubris — “here’s my idea,” and “the thing you don’t know is in the 1960s we invented that… since the 50s we’ve had the most brilliant people in the world working on these problems” — and overestimated addressable market, where the prize “might materialize in a single large contract, but a single large contract is not an enduring business.”

3. Drones are a monopoly game — and Lattis is how Anduril avoids playing it

• The structural math: “in every technology class in defense, there’s probably one, sometimes two actual programs that if you capture them, you have a business and if you don’t, you have no business… There are very few drone programs for small drones that would create a material enough amount of revenue to actually create a real business. You’re basically shooting the moon. You have to create a monopoly.” Harry’s obvious follow-up — why are there so many drone companies then? — gets a shrug: “there’ll be one that wins… the challenge for investors is actually figuring out which one it is.”
• Anduril’s answer was to go “really wide really quickly” via Lattis, “fundamental tech that allows you to consume data, make sense of it, and then manipulate robots basically” — a software platform branching into 20 different P&Ls. The proof of reuse: the 2017 sensing tower, built on “old school CV AI… pre cool AI,” shares code blocks with today’s autonomous jet fighter, giving each new market entry “a head start, usually reduced cost, usually a better schedule.”
• The gating rule for entering any market: “we wouldn’t get into a market if we didn’t think we already had applicable technology or a technology tailwind.”

4. 600 contracts, 20 that matter: how the machine allocates capital

• The hardest thing about government business isn’t the wins, it’s the gaps: “we’ll do 600 separate contracts this year. Only 20 of those are probably of material revenue size… what happens in between the 20, and what are you doing to position yourself for the 21st?” Predicting an unpredictable customer means blending budget rhetoric, warfighting theory, and technology trends into bets fielding “5 to 7 years from now — no one is saying today this is what I need.”
• New ideas start as small tiger teams spending internal IRAD, throwing “a dart” at the market: “you know you have something when the customer’s excitement starts to match your own and you’re walking down a path together as opposed to this weird vendor-customer relationship.” These efforts can often be characterized by “a singular champion on the other side of the table” — the same heroism thesis he credits to his old Palantir boss (likely Shyam Sankar). From there it’s “lily pad” jumping, contract to contract, toward “something very large.”
• Scaling decisions run through an internal investment committee — “here’s what we did, here’s what we discovered, here are the gates, here’s what it’ll cost” — and the edge is execution speed after conviction: “when we say go, we have the ability to do it faster and at a higher level of spend than most of our peers.”

5. Missiles built like bathtubs — the thesis the Iran war confirmed

• The most surprising upside bet: the Barracuda cruise missile family, entered years ago on a warfighting-gap thesis — low cost, ubiquity, “a mix of exquisite systems and mass,” commercial supply chains over aerospace ones. Then “there’s a war going on with Iran… a large percentage of our offensive and defensive missile systems have been expended and they’re really hard to replace.” The thesis was “confirmed, but confirmed in such a way where it feels wholly different.”
• The manufacturing insight: “the actual airframe is built like you build a bathtub — we can take advantage of tons of contract manufacturers in the United States… you can handle elasticity of demand for the first time. That’s not really a thing that’s ever existed in the missiles business. We can turn lines on and off.”
• On economics: “as a business we run 40%+ gross margin across the board — which in hardware and defense is quite good.” The counterintuitive spectrum: things you make thousands of run lower-margin, which “mostly comes down to customer expectation for what these things should cost over time” and the ability to regain margin through cost savings. In the Middle East today Anduril is “for the most part a defensive company” — one of the principal systems defending against the Shahed drone; on whether its drones were used in Iran: “I can’t speak to specific weapon systems used in specific situations.”

6. The misses: cyber regret, and a $40M scrap that bought the jet fighter

• The regret: offensive cyber warfare, now going public across the US and NATO after being “the spookiest of all spooky things… We should have moved into that 7 years ago and been the first in it. We weren’t big enough probably to really even understand it — but in hindsight it was kind of obvious.” Cyber is dangerous precisely because it’s asymmetric and non-kinetic: “when something explodes, we react to it with the full force of the military. When things are simmering beneath the surface in a cold-war-ish way, it becomes really hard to react” — attribution, escalation doctrine, and whether the US “can amass the same kind of force non-kinetically” remain publicly undebated.
• Harry’s pushback — with a $20B contract and Lockheed’s ~$100B revenue, why bother catching up? — gets answered by the platform logic: Anduril only enters where it already has applicable technology, and where it does, “we’re definitely going to go after it.”
• The mistake in the other direction: an early “pretty sophisticated aerial system” scrapped “20 million or 40 million in” — no core building blocks, no core team, no idea how the customer would buy it. His refusal to disown it: “do we even get to build the autonomous jet fighter if we had never made that mistake? I don’t think so… it’s hard to separate these failed things from what came after.” Same stoicism on unmade decisions: “I deeply learn from indecision and from bad decisions, but it doesn’t haunt me.”

7. Why Anduril wants to go public — and why it isn’t raising to buy VC-funded companies

• The IPO rationale is trust: “there’s a pedigree to it… no one would ever say it out loud, but when you are a public company, there is an additional level of trust afforded to you that you don’t have when you’re private.” Timing: “still a couple years out, but we’re close. We’re in the window” — with heavy spend already going into internal systems readiness.
• The operational gate: of 20 products, “only about a quarter are in any kind of rate production and throwing money back to the business” — the rest are losing money in J-curves Anduril runs in 3-5 years vs the industry’s 7-10 (Roadrunner: “24 months from a back-of-the-napkin drawing to a fielded system,” easily $100M+ before revenue). Losers get killed “before they ever get into high spend — don’t even enter the curve if you don’t have faith it’s coming through the other end.”
• On why not raise billions at the reported ~$60B and buy these companies: “it’s not the company we want to run. We want to create an enduring public company… We don’t have the luxury of making mistakes — our really big competitors, if they make a mistake, they get paid to fix it. If we make a mistake, we lose money and we lose contracts. We want to be a company in 2050 and 2060.” On multiples, he claims Anduril sits at “10 or 14x forward” vs deals at “20, 30, 40x” — Harry: “why don’t you times that by five, dude? I see like 200.” Steckman: “Great. It’s even worse.” His unlimited-checkbook target: space domain, where “the government wants to move really fast” but sits in a providerless gap between slow-but-exquisite legacy primes and SpaceX’s commercial adjacency — “a huge open hole in the market that we definitely would run at.”

8. Ethics, war profits, and the long horizon

• Asked whether there’s ever an ethical question about “sending thousands of missiles to Iran,” Steckman’s answer is clear: “we have a democratically elected set of governments that represent the US, the alliance, NATO… They set the rules and we abide by those rules. If you start to mess around with anything within that framework, you create a pretty dramatic set of slippery slopes.” The self-selection test: building missile systems that might only field in 2035 means “if you fundamentally lack trust in democratic institutions, this is not the game and this is not the business for you.”
• Is war great for business? “That’s the wrong view… over multiple decades everything kind of nets out. Many companies are profiting off the war in Ukraine — if those companies are myopically focused on that conflict, they will not have a business coming out the other end.” Anduril itself was “a couple hundred people” when Ukraine started — “you go to war with what you have… you do not invent new stuff along the way” — versus its heavy defensive role in the Middle East today as “a wholly different company” of ~8,000.
• Quickfire convictions: every current military mission “can be replaced with an autonomous system… as time goes to infinity” — though “we’re not even close to where a lot of people think we are on it.” And on Palmer’s brand: “I do unequivocally” think it helps — “in defense, every single thing is a headwind… Palmer is an N of one” when it comes to establishing trust relationships with people who matter. His own self-description, after Harry calls him unusually humble for an Anduril founder: “Never forget I am an international arms dealer.”