WORLD.MINDS Dispatch
June 2025 | Deep Dive on Drones
Written by Dan Steinhart
“Drones are the biggest disruption to warfare since Genghis Khan put stirrups on horses.”
—Erik Prince, founder of private defense firm Blackwater
Getting up to speed
There are 3 things to know to get up to speed on drone warfare:
- Everyone knows how disruptive drones are.
- Nobody knows what to do about them yet.
- Innovators are in an all-out sprint to solve the asymmetrical problems drones present.
You are now equipped to explore this Deep Dive, which is all about the companies racing to build drone and counter-drone systems.
Unfortunately, the western world has some catching up to do. For years, China was the only real innovator in drone hardware. I’d say we were asleep at the wheel, but in reality, US regulations stifled the industry. Now, spurred by the urgency of war, ten years of technological progress are being compressed into one.
Innovators in Ukraine and Russia are leading the way. But a motivated talent pool with plenty of capital in the US and elsewhere is sprinting to build next-gen drone tech. That includes AI swarms, microwave weapons, lasers, automated turrets, electronic warfare systems and American factories that pump out defensive drones by the thousands.
As you read, keep in mind what Peter Webb, former head of counter-drone operations at Anduril, told me on the Rational Optimist Podcast. I asked Peter: What’s the most important thing nobody understands about drone warfare? He answered:
The rate of change. What worked a week ago may not work tomorrow. There are people on both sides continually innovating because their life literally depends on it.
With apologies to the innovators building seaborne and ground drones, this Deep Dive will focus only on drones of the flying variety. Saronic Technologies – the rapidly growing maker of uncrewed, AI-piloted drone boats which recently achieved a $4 billion valuation – deserves its own article. As does Overland AI, whose new fully autonomous “ground drone” can drive 35 mph over difficult terrain with no human involvement.
First, let’s quickly look at why drones are so disruptive in the first place.
Then we’ll go through the drone and anti-drone technologies being developed, along with the key companies and innovators pushing them forward.
Asymmetric Warfare 101
“Asymmetric” comes up a lot in drone discussions. It means drones help the weaker side in war. Drones are the #1 reason Ukraine’s managed to stalemate its stronger enemy.
Ukraine deployed an astonishing 1.2 million drones to its armed forces in 2024 alone. It will build 2 million more this year. Drones are cheap, disposable and can be quickly produced and deployed in large quantities. But that’s not all.
Drones have “democratized” precision-attack capabilities. A cheap kamikaze drone can be precisely steered into the weakest spot of an individual tank, for example. A strike that precise used to take multimillion-dollar weapons systems. Of all the armored vehicles destroyed in Ukraine, drones were responsible for 70%. That’s despite drones failing to reach their targets 60-80% of the time.
Both Ukraine and Russia are leveraging drones. The US sent 31 of its finest M1 Abrams tanks to Ukraine, costing $10 million apiece. Nineteen have been destroyed or disabled, many taken out by Russian drones costing under $1,000, causing the rest of the tanks to be pulled off the front lines.
For perspective, the US deployed 1,956 tanks in the Gulf War. Enemy fire didn’t destroy a single one.
Former Green Beret Scott Zolendziewski, now head of growth at anti-drone startup CX2, told me during our meeting: “Drones are fundamentally an offensive weapon.” Boy are they ever.
Last year Iran launched a barrage of 350 drones and missiles at Israel. Israel’s advanced defensive systems thwarted the attack with a 99% success rate. But it was a good news/bad news situation.
Iran’s attack, based around its Shahed-136 drones, cost around $100 million. Meanwhile Israel’s Arrow interception system costs about $3.5 million per shot. Even Iron Dome, the cheapest option, costs about $100,000 per shot. In total, Israel spent over $1 billion to defend itself – more than 10 times Iran’s attack cost.
That’s the problem laid bare: you can’t keep spending millions to shoot down drones that cost tens of thousands of dollars. Which helps to explain why Israel is now investing heavily in RAFAEL’s Iron Beam, a laser-system that can disable a drone for just a few dollars.
It’s also worth noting that Hamas extensively used cheap drones to carry out the October 7 terror attacks. Drones disabled Israeli surveillance towers, dropped explosives on tanks and targeted guard posts. Although I am no tactical expert, it seems like the attack never could’ve happened without drones.
Cheap SMALL drones = BIG disruption
Cheap SMALL drones =BIG disruption
Drones in war aren’t new. The US has used them since Vietnam, and it still retains a big lead in making and using so-called “exquisite” drone systems.
You hear the word “exquisite” a lot talking to defense tech experts. It’s not a good thing, for reasons I’ll explain.
The US military’s drones are big, powerful, sophisticated and expensive. Like the MQ-9 Reaper. Price tag: $32 million fully equipped. Or the RQ-4 Global Hawk by Northrop Grumman, which costs upwards of $130 million.
These top-of-the-line drones can fly the fastest, highest and furthest. They’re equipped with the most sophisticated communication systems. They are the best military drones.
These are NOT the types of drones wreaking havoc in Ukraine. It’s the cheap, mass producible types that any ragtag militia can obtain that are redefining warfare.
These are called consumer or commercial drones. Their key trait: roughly 30,000 of them can be manufactured for the price of one MQ-9 Reaper.

The State of the Drone Market
(i.e. how the US handed China the drone market on a silver platter)
In 2015, China launched its “Made in China 2025” initiative. It was meant to build China into an industrial power in 10 years. Spoiler alert: it succeeded.
As part of the initiative, the CCP selected four “national champion” companies to lift up and subsidize. Drone maker DJI was one of them, along with BYD, Huawei, and Autel.
DJI was the most successful of the four. That’s really saying something, because BYD makes the world’s best electric cars. Better and cheaper than Tesla, I’m told by friends who’ve ridden in both. I’ve never seen a BYD in America, but they are everywhere in Europe.
BYD at least has a legitimate rival in Tesla. DJI has no rival. Chinese company DJI owns 70% of the global drone market and 80% of the US drone market (!). DJI dominates the drone market far more than Apple dominates the smartphone market.
And it’s not that DJI merely makes the most drones while US companies make the best commercial drones. DJI’s drones are the gold standard for hardware. They dominate in both quantity and quality.
How did we get to this sorry state? Well, China dumped cheap and subsidized electronics of all types, including drones, on the US in the mid-2010s. This crowded out US drone competitors.
American innovators probably would’ve put up a fight if they’d been allowed to compete on equal footing. We’ll never know, because America crippled its own drone industry with regulations. For example, they required a pilot’s license to fly a drone beyond line of sight, which prevented a consumer drone market from developing.
Meanwhile China classified drones as toys, allowing for rapid innovation. A consumer market blossomed. DJI’s drones followed the path of TVs: quality skyrocketed as costs plunged. DJI drones are now used in the US and elsewhere for search and rescue, filming, roof inspection and dozens of other tasks.
As the cherry on top, DJI received large investments from American venture capital firms Sequoia and Accell in 2014 and 2015.
China is now far and away the #1 drone maker. It will produce 8 million drones this year. Aside from Ukraine and Russia, everyone else only makes about 100,000.

More drones are shot down in Ukraine every day than the entire US Marine Corps has in its inventory. Ryan Tseng, CEO of Shield AI, an impressive autonomous drone company we’ll talk about momentarily, sums it up: “The industrial capacity of China is fearsome.”
That fearsome capacity stems mostly from one city: Shenzhen, the world’s factory. Inc.com reports 90% of the world’s electronic devices are made in Shenzhen.
DJI’s factories are so highly automated, drones can fly themselves from one station to another during assembly!
So if it’s purely a numbers game, China wins. Thankfully it’s not. Software and AI can level the playing field.
Autonomous Swarms
Nightmarish, but probably 5 years away
Not-so-coincidentally, Shenzhen recently hosted a Guiness World Record-setting drone lightshow. 10,200 drones danced together in the dark sky, shifting between patriotic Chinese shapes. It reminded me of the 2008 Olympic opening ceremony in Beijing, when 2,000 Chinese drummers played in perfect sync.

One computer controlled all 10,000 drones. It’s a bit terrifying. It doesn’t take much imagination to think what these drones could do if armed with bombs or guns. Worried? You’re in good company. “Autonomous swarms” keep generals and innovators up at night.
Contrary to popular belief, these light shows are not proof that killer autonomous swarms are at our doorstep. These simple drones are preprogrammed to follow a script. They don’t make decisions.
We’re still firmly in the “1 vs. 1” era of drone warfare. Generally, one drone is piloted by at least one human operator. Or sometimes a team of human operators – it can take up to 4 people to complete a drone mission.
Certain individual drones have sophisticated autonomous capabilities. Like Shield AI’s Nova, which we’ll discuss. But AI swarms are a nascent technology.
Eventually, a small number of humans will control a large number of drones that all coordinate with each other. And eventually, the swarms could function with no human involvement at all. Think 100,000 AI attack drones overwhelming an aircraft carrier or a city.
Militaries are preparing for this. When might autonomous swarms arrive?
Some Ukrainian officials predict autonomous swarms will debut in battle in 2025. Based on my conversations, I doubt it.
Former Green Beret Scott Zolendziewski from CX2 says: “Probably within 5 years swarms are going to be an issue. And if you’re playing on team defense, it’s going to be even more incredibly challenging.”
Peter Webb, formerly head of counter drone operations at Anduril, says a military could probably field an autonomous drone swarm today as a one-off. But the tech is nowhere near scalable, efficient or reliable enough for repeated usage.
But everyone agrees swarms are coming. That’s the bad news.
Now let’s cross to the sunnier side of the street and look at the anti-drone defense technologies innovators are aggressively working on. As you’ll read, many different methods are in rapid iteration.
This will not be a winner take all market. A layered defense, likely including each of the technologies below, will be needed to protect against autonomous swarms.
High Power Microwave Technology
Microwave weapons trace back to the Cold War. Their premise is enticing: shoot a burst of electromagnetic energy that fries the circuits of any electronic device in their path. This instantly disables the drone, or plane, or boat… or anything electronic.
Unlike bullets or lasers, microwaves can easily cover a wide area, disabling a whole swarm of drones at once. Microwave weapons can also penetrate fog and rain, where lasers struggle.
Sounds great, but shaky technology has long held HPMs back. Unti recently, HPMs were based on vacuum tube tech, which is fragile, bulky and immobile.
Epirus, headquartered in Torrance, CA, has changed the game. It was first to ditch vacuum tubes and instead integrate Gallium nitride (GaN) semiconductors. Like many recent innovations, this was only possible because of the huge improvement in computer chips.
Epirus’ flagship product is Leonidas. Its GaN emitter amplifies microwave pulses, condensing a huge amount of energy into a 1/10,000th of a second intense burst. The result is a compact, flexible and robust microwave weapon.

Epirus is an uncommonly fast-moving company. The defense “primes” had been working on HPM weapons for over 30 years. Epirus took Leonidas from concept to field testing in less than 7 years.
(The “primes” are the giant defense contractors that’ve dominated in the US for decades: Lockheed Martin, Northrop Grumman, Raytheon, Boeing, and General Dynamics, with L3 Technologies sometimes included.)
Then, after Epirus was awarded a $66 million Army contract in January 2023, it handed over 4 complete systems within 9 months. This speed is pretty much unheard of in the military procurement world plagued with delays.
Epirus is named after the fictional bow of Theseus, which had infinite arrows. Each microwave costs only a burst of electricity. The only “ammunition” is kilowatt-hours.
Earlier this year, Epirus secured an additional $250 million in funding. Its valuation is above $1 billion, making it one of only 9 defense tech “unicorns” worth >$1 billion. The company is using this money to scale. It can currently produce 20-30 Leonidas systems a year. It wants to produce 100 or more.
Figures aren’t disclosed, but one Leonidas system likely costs $15-20 million. Although that’s a significant upfront investment, Leonidas earns it back in “Cost per Engagement.” In testing, Leonidas was able to take down 100 drones in one shot.
Co-founder Joe Lonsdale says it can shoot miles, although exactly how far is a secret. He sees Epirus’ weapons one day guarding stadiums and airports. He also says the technology can be shrunk down to be placed in the head of an Anduril Roadrunner drone, for example, perhaps to provide protection for any large gathering, political event, and so on.
An aside about Lonsdale: he’s the best defense tech investor. Of the 9 defense tech “unicorns,” he either co-founded or was an early investor in 6:
- Epirus
- Palantir – the pioneering “neo prime,” now worth $139 billion
- Anduril – more below
- Shield AI – more below
- Saronic Technologies – drone boats mentioned earlier
- Chaos Industries – next gen sensor technology.
Lonsdale is a venture capitalist, so he’s surely had his share of failures as well. But man, what a list. If you could only pick one defense tech investor to follow, he’s it.
Epirus CEO Andy Lowery often compares Leonidas to a forcefield. The company is now working on “Leonidas H2O,” a variant that disables boat motors and was demoed for the US Navy in August 2024.
Leonidas’ most prominent competitor is Raytheon’s PHASER, which is based on older vacuum tube technology. It sits on a shipping container, which gives certain logistical advantages:

Some say the older vacuum tube technology has advantages. It can reach a higher peak power. And militaries have been working with it for longer, so it is more reliable. As a rational optimist, I try to keep the cynicism to a minimum. But it sounds like Raytheon has already sunk so much money into their older tech, they’re unwilling to change.
The other notable competitor is THOR, which the US Air Force Research Lab is developing. It too is based on older vacuum tube technology. But it differentiates itself with an ultra-wide beam that can cover a football field-sized area.
THOR is years behind Epirus. Considering how fast Epirus moves, it likely always will be.
HPMs have drawbacks. For one, drones can be shielded against HPM blasts, rendering microwaves useless. But doing so is expensive and adds weight to the drone, undermining the main advantages of drones.
It will probably get cheaper and more efficient to shield drones against HPMs, simply because that’s what innovation tends to do. The swarms of the future could be a mix of bigger, more expensive drones shielded against HPMs, plus fleets of tiny cheap drones. We’ll need a layered defense.
Also, Epirus is a big, expensive system, an obvious target for drones attacking it from all angles. It will need to be defended, perhaps by automated guns… which brings us to our next category.
AI Turrets
Turns regular guns into anti-drone AI superguns
This fast-developing category melds regular guns with state-of-the-art AI tech.
It’s only a slight oversimplification to say these turrets turn a standard military machine gun into an AI-controlled, superhumanly fast and accurate gun. These systems are much smaller, cheaper, and nimbler than HPM systems. They can passively track the sky, lock onto a drone, and await “destroy” orders from the soldier.
Allen Control Systems (ACS) is a fast mover in the US. It recently raised $30 million in its series A round. Its flagship product, a turret called Bullfrog, weighs 165 pounds and can be mounted on a truck.

Bullfrog takes an M240 machine gun and imbues it with computer vision and AI, allowing it to identify and aim at enemy drones. If the solider wants to shoot down that drone, he presses one button. Bullfrog is capable of performing the kill autonomously too, without any human involvement whatsoever.
However, the military is reluctant to allow automated weapons to fire without a human making the kill decisions, for ethical, safety and public-perception reasons. Expect serious debate on this as autonomous swarms approach viability.
ACS doesn’t disclose the cost of a Bullfrog system. But the company publicizes a cost-per-drone-kill as low as $10, and a maximum effective range of 1500m, or almost one mile.
Bullfrog has performed well in tests and markets itself as effective against swarms compared to other AI guns. In a recent demo in California, it defeated a swarm of 7 drones. The company also emphasizes that Bullfrog is easy and quick to deploy. A soldier can be trained to use one in less than 30 minutes.
The Israeli firm Smart Shooter is a key competitor. It makes hardware that turns regular guns into ‘smart guns’ equipped with AI that identifies and locks on targets. Its products primarily attach to handheld guns.

Smart Shooter was founded in 2011 and has recently pivoted to anti-drone tech. Its SMASH Hopper station is basically a ‘lite’ version of the Bullfrog, weighing only 35 lbs. Unlike Bullfrog, SMASH Hopper needs additional external sensors to detect faraway drones. Earlier this year the company debuted the SMASH DOME, a full anti-drone track/target/destroy system that incorporates the Smash Hopper.
From a pure tech perspective, ACS seems better positioned in counter-drone defense. Its Bullfrog has the major advantage of being able to detect/identify/defeat a drone from a self-contained turret, making it more mobile and robust.
Smart Shooter, however, is further along as a company. It is already a trusted supplier to western militaries. SMASH Hopper has been field tested and deployed by Israel, the US, UK, Australia, and India. This advantage in distribution and relationships makes it a serious competitor.
Honorable mentions: the enhanced gun space is getting crowded fast. In the US, AimLock makes a semi-autonomous remote weapon station. ZeroMark focuses on augmenting existing rifles with AI aiming assistance. Hover, Australia’s Electro Optic Systems, and Ukraine’s Dron ZP are players too.
Compared to HPMs or lasers, AI turret technology is straightforward and inexpensive. It’s easy to imagine every other military truck being equipped with an AI turret. They’re a solid solution to small drone attacks, but swarms could overwhelmed them.
Lasers
Melt holes in drones, but quite hard to build.
Lasers seem like a natural fit for cheaply shooting down drones. Like HPMs, their ‘ammunition’ is electricity, and they can shoot drones from more than a mile away.
Most of the innovation in lasers is confined to the giant military contractors. Building lasers powerful enough to take down a drone is complex and requires a lot of upfront investment. Lockheed Martin, Raytheon, Thales in France, and RAFAEL in Israel dominate the space.
RAFAEL is leading the development of Israel’s Iron Beam, the most advanced anti-drone laser system. It uses 100-kilowatt high energy lasers to disable threats, including drones, up to about 6 miles away. It only costs a few dollars per shot to operate, but the system took a whopping $1.7 billion investment to build. It is on track for deployment by the end of the year.
One smaller company has managed to break through in lasers: BlueHalo. Its flagship LOCUST uses radar to lock on to a target, then melts a hole in a drone’s vulnerable spot. It has been used by the US Army since 2022.
BlueHalo was acquired for $4.1 billion in May 2025 by AeroVironment, the most valuable publicly-traded military drone company. AeroVironment is best known for its “loitering munition” Switchblade drones, which Ukraine has used extensively. They were also the first drone company to earn a contract from the Pentagon’s Replicator initiative, meant to accelerate the US military’s adoption of autonomous technologies.
AeroVironment is a highly-respected and formidable company, especially with its acquisition of BlueHalo. But it’s focused on conventional military-use drones rather than the disruptive drone innovation I’m highlighting in this Deep Dive.
Honorable mention: Aurelius Systems in San Francisco. Aims to make miniaturized lasers systems that can go on any small vehicle, like a Humvee. Aurelius is very early-stage. It was founded in 2024 and has 9 employees.

Electronic Warfare
Drones are useless if they can’t be controlled.
For swarm defense, electronic warfare is the one category that currently gives HPMs a run for their money.
Imagine a battlefield where the air is weaponized. Electronic communications are strangled. Signal jammers lead drones and missiles astray.
This is modern electronic warfare (EW). It’s a fast-moving game of cat and mouse. In the early days of the war, Ukraine’s drones cut right through Russia’s unprepared defenses. But Russia quickly adapted, deploying thousands of jammers. These systems create bubbles that sever drone video feeds, block GPS and disrupt communications.
Every fourth soldier in Ukraine now carries a “jammer.” They’ve become as essential as rifles. Jammers can scramble the communications of drones, rendering them useless.

The largest jammer company in the world is Unwave, in Ukraine. It produces 5,000 jammers a month. Both sides now deploy handheld jammers – here’s an image of a Russian model.
CX2 is a newer American EW company that recently came out of stealth mode. Its goal: deliver “spectrum dominance” to the US and its allies on the battlefield.
I spent two hours with co-founder Nathan Mintz (also a co-founder of Epirus) in his El Segundo office/warehouse recently. I held a test drone – they are surprisingly flimsy. And I watched 3D printers whir, creating custom drone parts.
I may be biased, but talking to Nathan feels you’re talking to the world’s #1 expert on practical electronic warfare. Before Epirus, he spent 14 years designing EW systems for F-15s and F-18s at the primes. His desire to start CX2 came after an alarming call with his now co-founder, who had seen the revolution in drone warfare firsthand in Ukraine. That conversation prompted Nathan to visit Ukraine to see it for himself.
In Nathan’s words: “Our vision is to give America an invisible electronic wall that we can push forward and pull back as needed, to dominate the spectrum.” This electronic wall would deny enemies the ability to operate any electronics, while giving its forces free reign.
CX2’s three initial products all work together in a kind of small defensive swarm:
Wraith does reconnaissance. It can loiter in the air for 40 minutes, hunting for enemy signals. The entire drone body serves as a sensor, creating heat maps of radio signals across the battlefield.
Banshee is the strike drone. Once Wraith finds an enemy signal source, Banshee can autonomously track and destroy it.
Vadris is a small sensor package. It bolts onto existing drones, allowing pilots to “see” radio emissions. It adds visual indicators that appear on screen, guiding them toward enemy systems like jammers.
EW is powerful, but it isn’t a trump card. “Signal hopping” can stay a step of ahead of jammers. “Tethered” drones, connected to a thin cable, are a common workaround for short-range attacks in Ukraine.
The biggest upcoming challenge for EW is dealing with autonomous drones. I’ve learned not to make absolute statements about this rapidly evolving area. But autonomous drones cannot be jammed in the conventional sense, because there’s nothing to jam. They don’t need a signal sent back to a pilot. They make decisions to carry out a mission on their own.
For this, Nathan says, it’s even more important to follow his “shoot the archer, not the arrow” strategy. In other words, find the source of the drones and eliminate it.
As far as competition, there are many, many EW companies out there. Many are in Ukraine, and it’s hard to know what’s really going on in the fog of war. But in all our research, we did not find a direct competitor to CX2. That is, a company whose sole mission is to dominate the spectrum, and makes a suite of drones to achieve that goal.
What I can confirm is no EW competitor can match CX2’s mix of pedigree, history of success (Nathan with Epirus) and backing (Joe Lonsdale’s 8VC is an investor). The company is still early stage, having just raised $31 million in its Series A round last month.
Shield AI
#1 autonomous drone company… and the favorite to figure out autonomous swarms first?
As dominant as China is in hardware, the US still makes the world’s best software. And San Diego-based Shield AI makes the best autonomous software for military drones.
Its flagship technology, Hivemind, allows drones to operate completely autonomously, even in challenging spots like inside buildings or underground. Being fully autonomous, it can operate in GPS-denied environments.
The US military already uses Hivemind for reconnaissance, surveillance and electronic warfare. It is considered the gold standard for autonomously navigating in combat environments, a step ahead of anyone else.
Shield AI’s Nova quadcopter was the first ever autonomous drone deployed by the US military in combat situations. It is purpose-built to for tight, urban and indoor environments. Its V-BAT drone can take off, land and hover vertically, then transition to high-speed horizontal flight. It is deployed by the US military in the Black Sea and the Middle East.
If I had to bet on a company getting to autonomous swarms first, it would be either Shield AI or Anduril. We will discuss Anduril momentarily. Shield AI’s “V-BAT Teams” innovation already lets a single human operator command multiple drones simultaneously. This is the infancy of coordinated swarms.
In addition to making its own drones, Shield AI partners with other companies who want to use Hivemind in their drones. The company has attracted top investors, including Andreessen Horowitz and Joe Lonsdale’s 8VC. It’s a well-established company with a $5.3 billion valuation.

Skydio
Top manufacturer of drones in the US, but it’s in the doghouse
Skydio has produced 45,000 drones – a drop in the bucket compared to China and Ukraine, but enough to make it the leading US drone manufacturer by quantity. Skydio is highly respected for its AI capabilities, specifically in obstacle avoidance and consumer-friendly usability.
Unfortunately, Skydio’s name is mud in the defense world. It sent over 1,000 drones to help Ukraine. Many dropped out of the sky, falling easy prey to Russian jammers. Beyond that, Ukraine’s soldiers complained of issues with control and take-off reliability.
To his credit, the CEO Adam Bry was open and honest about the failure. He was quoted in the WSJ saying Skydio’s drones were “not a very successful platform on the front lines.”
We probably haven’t heard the last of Skydio, but it must rehab its reputation. It no longer makes consumer drones and focuses instead on producing for businesses and government. Most recent valuation: $2.2 billion.

Neros
Scaling drone manufacturing to fight swarms with swarms
How do you beat a swarm of 100,000 autonomous drones?
Perhaps with a swarm of 1,000,000.
That’s the bet 22-year-old wunderkind Soren Monroe-Anderson, CEO of Neros, is making. He believes the only truly reliable way to beat a big swarm is with a bigger swarm. Unlike most prominent US drone companies, Neros is focused on hardware.
Monroe-Anderson points out that swarms aren’t just a software problem. In order to field swarms, the drones that comprise them must be cheap and efficiently produced. Western countries, with the exception of Ukraine, lack the ability to do this.
Most of the electronics that comprise drones – batteries, sensors, cameras, motors – are made in Shenzhen. When China sanctioned Skydio in late 2024, it was forced to take the “drastic step of rationing batteries to one per drone.” Those are the Skydio CEO’s own words.
So Neros isn’t just making drones; it’s making factories that pump out ultra-low cost and lightweight war drones. With its first factory under construction in El Segundo, Neros aims to be able to make hundreds of thousands of drone systems a month.
Neros prioritizes low cost and light weight. Its Archer drone is tiny. It weighs about 3 lbs and is the size of a dinner plate. Archers are stackable, so they can be shipped in large quantities easily. To launch an Archer, simply toss it into the air.
Crucially, Archer is made entirely from non-Chinese components.
Neros is small, having most recently raised $35 million in its series A round. It faces a tough road. Scaling manufacturing is hard and expensive. Any company trying to solve this particular problem risks being the pioneer who gets riddled with arrows. We should all be grateful for Neros taking on the challenge.

Helsing
Europe’s top (and only) major defense innovator
Helsing is the company we’d get if Neros and Shield AI had a European baby.
Helsing is Europe’s biggest and best defense tech innovator. Its main product, a software called Altra, is comparable to Shield AI’s Hivemind. Helsing is the go-to AI software provider for European militaries.
Europe has long lacked defense innovators. This is now an urgent problem as European governments seek to avoid relying solely on US companies for defense. Helsing is perfectly positioned to take advantage. It has local expertise in Europe’s many regulations, crucial for doing business with governments there. It also has close ties with Germany’s military.
Helsing is foremost a software company, but it has now expanded into hardware. The HX-2, a small autonomous attack drone weighing about 26 lbs, is its first product. It has demonstrated some early success with swarms. Multiple HX-2s can coordinate through its Altra platform to form a small swarm.
Like Neros, Helsing aims to build a self-sufficient supply chain to eliminate reliance on China. Its first “resilience factory” in Southern Germany can churn out 1,000 drones a month, having only come online in late 2024. Its second plant is in the planning phase.
Helsing’s drones have faced some criticism in Ukraine. Some soldiers say its attack drones are more expensive and less effective than comparable products.
As $5.6 billion, Helsing is the most valuable European tech defense company by far. Its closest competitors are Quantum Systems in Germany and Tekever in Portugal, each worth around $1 billion.
Anduril
The neo prime Pioneer
With a $28 billion valuation, Anduril is the largest non-publicly-traded defense tech company. Compared to the primes’ combined market capitalization of $550+ billion, it is still quite small.
Despite its size, in my opinion Anduril is already one of America’s 10 most important companies. No other company besides Palantir has done as much to drag western militaries forward into our tech-first future.
We know the US still leads in producing “exquisite systems.” These are big, expensive “Queen of the chessboard” military assets, like aircraft carriers. The US has 13 carriers. The most any other country has in operation is two.
But the next era of asymmetric warfare is here, and the primes have failed to keep western militaries a step ahead. That was bound to happen with their perverse incentive structure. The “cost-plus” model guarantees a prime will be paid its costs, plus a profit, no matter the ultimate cost of a project. So they build slow and fail to innovate. What’s good for their shareholders is bad for the country.
Palantir and Anduril broke the stranglehold the primes had on military contracting. Anduril in particular flipped the cost-plus model. It frontloads R&D spending to develop products and then sell them, like a Silicon Valley company. This is risky and requires long-term investors who believe in the company’s mission. Anduril has them in Lonsdale’s’ 8VC, Peter Thiel’s Founders Fund, Mark Andreesen’s a16z and others.
Anduril’s foundational product is Lattice, an AI software platform that integrates data from cameras, radars, and other sensors and systems. It then parses this information with AI, turning it into useful insights to help soldiers make decisions.
Lattice also powers Anduril’s autonomous capabilities, which are just as formidable as Shield AI’s. Whereas Shield AI is best for navigating tight spots, Anduril is best at coordinating bigger missions. To rescue a hostage, choose Shield AI. To coordinate a dozen drones performing a reconnaissance mission, choose Anduril.
Anduril is far more than a drone company, but its lineup of drone and anti-drone tech is extensive. It has landed several big contracts, including a $600m+ one with the Marine Corps to build AI-powered anti-drone systems to protect military installations.
A sampling of Anduril’s drone products:
Anvil Interceptor: Quadcopter-style drone that autonomously detects and intercepts hostile drones by colliding with them.
Roadrunner Interceptor: A missile/drone hybrid that takes off like a helicopter but flies like a jet. It autonomously identifies and engages aerial threats, returning to base if engagement is unnecessary.

Pulsar: A box no bigger than a toaster that detects and disrupts enemy drones. Watch this 2-minute marketing video of Pulsar disabling an entire swarm of 75 drones in one pulse. The drones stop mid-flight and drop like dead flies, quite literally. The demo is real –Anduril has a strict “no CGI” policy.
Fury: A “loyal wingman” drone, designed to operate autonomously alongside crewed fighter jets.
Ghost: Small, lower-cost autonomous tactical drones that can be built and deployed in large numbers. One solider can manage multiple Ghost drones, a precursor to swarms.
The key word to describe Anduril’s overall strategy is scalable. It isn’t just inventing and designing better equipment. It is productizing autonomous military equipment in order to produce it much faster and cheaper than is possible today.
That’s why Anduril is investing nearly $1 billion to build Arsenal-1, a “hyperscale” manufacturing plant in Ohio. Arsenal-1 will mass-produce tens of thousands autonomous military systems annually, including drones.

Ukraine
The Silicon Valley of drones, shrouded in the fog of war
It feels a bit wrong to write about Ukraine’s drone companies with war ongoing, and it’s hard to know with much confidence what’s really going on due to the fog of war, secrecy and rapid iteration.
But we must at least mention the Silicon Valley of Drones: Ukraine. Ukraine’s drone production and innovation have skyrocketed out necessity. The country went from making virtually zero combat drones in 2022 to making 2 million annually today.
Some specific accomplishments:
- TAF Drones scaled production incredibly fast, now cranking out around 40,000 per month from its 20 production facilities. Despite only being founded in 2022, it can produce as many drones in a month as the leading the US company, Skydio, has made total.
- Skyfall was the first Ukrainian drone maker approved as a supplier by the US Department of Defense. Its drones are world class at withstanding jamming. Skyfall is now partnered with CX2.
- Prominent nonprofit group Wild Hornets makes the STING interceptor drone. It was one of the first inexpensive drones that could reliably destroy much more expensive attacking drones.
Vyriy and Ukrspecsystems are major players too.
Admittedly, this is pure speculation. But when the war is over, I expect Ukranian drone companies to be acquired for hundreds of millions, if not billions of dollars. Their manufacturing knowhow alone is extremely valuable, seeing as how several Ukranian companies exceed anything any other western country can do.
Finally, in the interest of completeness… similarly rapid innovation is happening on the Russian side.
The End?
Nope
Although the drone industry has skyrocketed in importance, it is still surprisingly tiny. Even mighty DJI is estimated to be worth only $15-30 billion, comparable to Domino’s Pizza. While no one knows the future, it is close to certain this industry will grow very fast.
What’s not a certainty is that the innovators will win. Despite all the money and brainpower pouring into the drone space, there are entrenched interests who don’t want to see the primes disrupted.
Here’s the skeptical take, summed up by world-class military expert and Blackwater founder Eric Prince. This was from a private conference in May, in conversation with my Rational Optimist Society co-founder Stephen McBride:
“I’m happy to see a lot more defense tech investing going on, a lot of VC-type firms now pushing into defense tech. Great. There’ll be a lot of great innovation…
…but it’s not going to go much farther than that because the top five defense firms occupy such a strategic ground of lobbying in Washington, which influences Congress to buy certain high-dollar exquisite programs versus the small underdog.”
That said, I’m not convinced the giant US military primes will be able to hold off the drone innovators. The primes are infamous for notoriously long and blindingly expensive procurement cycles. We cannot afford to be slow while our adversaries are sprinting.
I’ll leave you with a thought from CX2’s Nathan Mintz:
If, ten years from now, at least one or two of the big defense primes haven’t been fundamentally disrupted, America is in serious trouble.
About the Community Dispatch
This Deep Dive is published exclusively in collaboration with our friends from the Rational Optimist Society, which co-founded by our community member Matt Ridley. WORLD.MINDS members enjoy exclusive early access to their newsletter. They publish other insights, so sign up to their substack. The companies featured in this newsletter do not pay to be included. Neither WORLD.MINDS nor the Rational Optimist Society receives compensation. We highlight ideas and people we genuinely admire.