Munitions Made in Shenzhen

Clawing back drone manufacture

If you’ve watched a $500 drone turn a $5 million tank into a smoking hole and thought, somebody is getting rich on this, you’re right. It’s just nobody in the US.

Everyone already knows that drones are changing war, it’s been on the news every night for three years, but here’s the part nobody has said out loud: the country that invented precision airpower cannot currently build the cheap version of it. The reason for that is the stupidest you have ever heard.

From exquisite to expendable

For two decades, American airpower meant the MQ-1 Predator and MQ-9 Reaper — beautiful, expensive, MALE-class machines (Medium-Altitude, Long-Endurance) that cost tens of millions and that nobody but a superpower could field. In the trade these are called “exquisite” systems. Exquisite as in expensive, small in number, and precious. You do not throw one away to kill a truck.

Look at the Second Nagorno-Karabakh War in 2020. Azerbaijan took $1–2 million Turkish Bayraktar drones and ~$700k Israeli Harop loitering munitions and used them to find and gut Armenian armor. The Armenian separatist army, dug into its mountain defenses, had no real way to jam what was hunting it. Cheap eyes in the sky plus cheap fire from the sky, and a hundred years of armored-warfare doctrine got blown up.

Now consider Ukraine. The dominant weapon of that war is not a Reaper, but a first-person-view quadcopter, a hobby airframe with a warhead zip-tied to it. They run $200 to $2,000 a unit, built from commercial parts. They can be iterated weekly and flown straight into a tank that costs a thousand times more. The industry word for this is “attritable,” as in cheap enough to lose. If you send ten and one connects, you came out ahead by orders of magnitude. That’s the inverted cost-exchange ratio.

• FPV strike drone takes out armored vehicle: $200–$2,000 / $1–$5 million = 0.00004 - 0.002.

Now take the Geran-2. The Russians fire a lot of them into Ukraine the same way the Iranians fire Shaheds across the Strait of Hormuz. They’re kind of trash, but they’re carrying around a lot of bomb load right now in 2026. The American defense against them? Exquisite.

• Air-defense missile takes out Russian Geran-2: $3.5 million / $35,000 = 100.

Who is more likely to win by that arithmetic? Hint. Smaller numbers are good.

This isn't a fringe take. NATO's deputy supreme commander in Europe, Sir John Stringer, says the days of sitting back and engaging every threat with fast jets and surface-to-air missiles are over, and that swatting $35,000 Shahed-style drones with multimillion-dollar Patriot interceptors is unsustainable. War has stopped being about who has the most exquisite machine and started being about who can stamp out the most attritable ones. These are the munitions of the modern battlefield.

So, who has the advantage here? Who makes the cheap drones? Here is the part that should make your teeth itch.

We outlawed this industry

There is no magic in a $500 drone. Strip one down and it’s carbon fiber, aluminum, a little wood and plastic, a battery already perfected by the cell-phone industry, a motor you can wind out of ordinary copper, the same three-dollar control chip that Texas Instruments has been stamping out by the truckload since the 1990s. These are parts that any motivated teenager could find.

And American teenagers were. A decade ago, the hobbyist drone scene here was genuinely world-class. But then came the FAA. The regulatory apparatus met the emerging drone industry with a reflexive, risk-averse, process-bound “no.” You can’t do that, you can’t fly there, you can’t sell this. Fill out these forms, and wait until you give up. The hobbyists who could have launched a million hundred-dollar drones spent fifteen years wrestling with paperwork, by which point somebody else stepped in.

China, meanwhile, let its builders build. Now, they own the global supply chain. Look at the manufacturer’s label on any hobby drone, or on anything you order off Alibaba. When Ukraine needed drones at scale, a meaningful chunk of that supply ran through Shenzhen. When Beijing throttled that pipe in response to political pressure, Ukraine started losing ground. They clawed it back with home manufacture, kitchen-table assembly, and a frankly heroic gray-market smuggling operation routing parts through Europe and the Middle East. They’ve survived so far, but picture trying to win a war while your ammunition supplier is your adversary’s close friend.

If we’d had drones to sell to Ukraine … but we didn’t. We regulated ourselves out of the most important weapons category of the century.

Too Lean

The instinctive objection is, “it’s just cheap Chinese labor — structural cost advantage, nothing to be done.” Yes, China’s mass-manufacturing is a genuinely fine art and no, a domestic small-run operation is probably never going to beat Shenzhen on raw unit price. This is how China got ahead, but unit price was never the axis we’d win on.

Any cheap source manufacturing is always going to have an edge, but the thing you want is not always the cheapest thing. Everyone out there has to ask themselves whether their alliance with the PRC is permanent, and what happens if things go wrong and shipping gets interdicted.

For a generation, just-in-time manufacturing (and lean manufacturing overall) led us into horizontally integrated supply chains, sometimes of near-monopoly production. All Hard Drives Come From Thailand, Your Syringes Are Probably From The Philippines, Your Tires Were Made In Sri Lanka. It’s efficient, right up until it isn’t. A six-month chocolate shortage makes people bitchy until they get over it. A shortage of insulin, or penicillin, or bullets, and people fucking die. Drones just moved into that second category.

Nations have strategic imperatives for certain critical things necessary to maintain their sovereignty. In the modern world, that's things like food (and fertilizer), fuel/power, fresh water, and the ability to protect their borders and their citizens. In that last category you have both the literal armed forces and police and the munitions industry necessary to supply them. If you have weapons, but you’re dependent on someone else for the ammunition to keep fighting, you can’t defend yourself very long. This has led to things like concern about shortage of artillery shell manufacture, but also of course a dependency on all drones coming from China — a grave concern if you happen to be, for instance, Taiwanese.

Every nation that watched what the US did with the Predator and Reaper (and, more to the point, what Turkey did far more affordably with the Bayraktar) wants that kind of long-flight monitoring and strike, especially for policing vast expanses of ocean and low-intensity asymmetric threats like the pirates off Somalia or Malaysia. Poland is near the top of the list, and of course every Gulf state is concerned about Iranian (and Houthi) drones and speedboats. Sri Lanka wants cheap autonomous eyes over its coastal waters so badly it’s prototyped some of its own. Some nations won’t even try to build it — a manufacturing lead is hard to catch up on, and plenty of them will just stockpile drones, parts, or both, because they don’t believe they can stand up a local infrastructure. Europe is planning to spend up to €800 billion on defense by 2030, with drones threaded all the way through it. Everyone at once wants out from under the single point of failure.

Not everyone gets to be choosy. If you’re an African warlord with no industry to build your own drones, you shrug, take the bribes, stockpile what you can, and you hope the Chinese don’t decide to back your rival. You’d have taken the same deal from the Soviet Union, or the Cubans, or the CIA, or the French. China wins on price, but if you’re crosswise with Beijing’s aims, and you can imagine an insurgency suddenly enjoying Chinese support, you’d better have your own source of weapons. You need indigenous manufacture of the strategic imperatives, or a strongly allied partner who can’t be cut off by a casually sabotaged railroad. It’s wildly inefficient for most nations to run their own defense industry instead of shopping the international arms bazaar, but there are good reasons for them to do it anyway.

The right to build Bayraktars under license and make derivative versions thereof will be very useful; a whole lot of people don't trust the Turks but would like this class of drone because it's proven out. The Israelis have the same problem selling their drones; they mostly work, but a lot of people don't get along with Israel Military Industries and Israel Aerospace Industries, or can't be seen dealing with them. The appealing part for a lot of these potential customers is not just a US offering, but an indigenous solution: US tech / engineering, but with local control and manufacture so that they don’t have heartburn over their supply chain.

The suits are finally panicking

It’s fun to watch the official answer go from “no” to “yesterday.”

In June 2025, the Trump administration issued an executive order titled, with zero subtlety, “Unleashing American Drone Dominance”. A month later, the Secretary of Defense signed a memo reclassifying small drones as consumable assets — ammunition — handing procurement authority down to individual commanders, and announcing that the bureaucracy’s gloves were coming off.

Builders have been screaming for a decade that the real risk is risk-avoidance. Nice of everyone to catch up. By the end of 2025 the FCC had banned new foreign-made drones (read: DJI drones) from authorization for sale, and the War Department had put out a request to industry to build 300,000 drones, fast and cheap, backed by a roughly $1.1 billion program. The man who couldn’t get a hobby permit in 2015 is now the guy the Pentagon is begging to scale.

The market has, predictably, lost its mind. The names in this space — Red Cat (RCAT), Skydio, Anduril, Unusual Machines (UMAC), AeroVironment (AVAV), Palladyne AI (PDYN) — have ripped on every headline, some of them up several hundred percent on the year. One of them, Unusual Machines, counts the President’s eldest son as an advisor and one of its largest shareholders, which tells you everything about where the connected money thinks this is going.

The actual edge here isn’t even the $500 airframe — that’s a commodity, and commodities don’t have margins. The edge is everything bolted on top: the autonomy, the AI targeting and navigation that lets the thing find its mark through jamming and spoofing, the software layer where a $3 chip turns into a weapon. That is where we could lap the Chinese, or likewise where they will push the frontier. There is a stupid amount of interest within the FPV drone space both in the fiber optic drones and in how to combat them.

The frontier past that is autonomy. No, not full autonomy. Everyone always goes down the Terminator rabbit hole, but we don’t want to whole-hog on killer robots. What if they turn around and target our own side? What people are very interested in, though, is fire-and-forget: a drone you maneuver into range, “click on” a target, and let it handle its own approach and impact.

This isn’t hypothetical. Back in 2021, a UN panel reported that a Turkish STM Kargu-2 “hunted down and remotely engaged” retreating fighters in Libya with what the report called a true “fire, forget and find” capability. The panel never actually confirmed a kill, and the breathless John-Connor coverage oversold it, but the capability is real and shipping.

Here’s where the two-tier supply chain bites again. A toy FPV drone runs on a $40 flight-controller board; an autonomous killbot doing its own computer vision and path-planning needs real silicon — an NVIDIA Jetson-class companion computer, some of it under US export control. Off-the-shelf AI guidance modules are already for sale on Alibaba, but buyers don’t trust the PRC to not put a black box in the kill loop. So, the clone projects are already underway. Ultimately, Texas Instruments can make those chips too and Sony can make the cameras. Those are names more worthy of trust than DJI.

The credentialed system spent fifteen years telling the competent people to give up, and in doing so, it handed the decisive technology of the era to an enemy. Now, what’s the next inversion, and who’s already building it?

Learn to see it coming, because the institutions won’t.