This is a grainy, sped up, cut up, loud and proud combat highlight tape released to social media by a Ukrainian military brigade. In the picture are Russian T-72B3M tanks, a refurbished and revamped Cold War staple that Russia has come to rely on during their invasion of Ukraine. All three are doomed, about to be put out of commission by the author of this video, the 79th Air Assault Brigade—a highly trained but highly out-gunned subset of the Ukrainian defense effort. The unit deals with all three tanks through conventional means—weapons and approaches that have more or less been around since World War II. The first is stopped in its tracks by a mine, an anti-tank mine, while this one, unclear as to why it’s stopped, is hit by an antitank missile or some similar sort of air ordnance. Finally, what struck the second is what strikes and stops the

third. Up to this point, if the video were just a 
bit grainier or black and white, the battle scene   could be from 2024, or 1974, or 1944. 
But the clip isn’t over.  These three tanks—at least by their own 
power—are not going to move again, which   for the 79th, means they’ve landed three mobility 
kills, as they’re called in military parlance. But   the work’s not done—should they be hauled back 
behind Russian lines, perhaps they could be fixed,   or at least used for parts. So the 79th needs 
to finish the tanks off. But now they turn to   something novel, something that’s shaped 
Russia’s invasion and Ukraine’s defense,   something that’s quickly becoming a 
staple in 21st century warfare: drones.   Rather than sending troops in to finish off tanks 
and chase down their operators—which risk the

safety of a soldier—or fire artillery or missiles 
at the sitting targets—which the Ukrainian military has desperately few of—the 79th now 
mobilizes a fleet of small, inexpensive unmanned aerial vehicles. Here, a fixed-wing drone strapped 
with an explosive plows into the tank’s weak point to render it useless. Here, another does the same. 
Here, a drone pilot, rather than tracking on foot, follows the path of hiding Russian soldiers with 
what’s in all likelihood a quad-copter equipped with a grenade. Here, the same. And, when zooming 
out to consider what made this possible—spotting the quickly moving enemies approaching first, 
then filming it all to cut the prideful 1-minute, 27-second piece later—it’s again a drone, 
this of the more expensive military-grade variety. And these 90 seconds serve as a 
microcosm—drones are no small part of what’s kept

Ukraine in the fight for this long, and they’re a big part of why the 79th still exists at all. Consider the context of this clip—it’s shot somewhere around here on the front lines in the hotly contested Donbas region near Donetsk. For years now, the 79th Air Assault Brigade has been battling over this region—in particular over this village, Marinka. Once home to 9,000, Marinka is now a ghost town with few structures still standing. It’s a spot that Ukraine and Russia grappled over for 20 months before it finally fell to the invaders in late 2023. And to some educated onlookers and strategists, its fall was cause for concern. Now, it seemed Russia would have easier access to Ukraine’s interior by way of the city of Kurakhove, just 10 miles or 16 kilometers west down O0510 Road. But in more than 6 months of trying and 14

coordinated efforts to break through, Russia’s 
been rebuked—because of brave, well-trained soldiers, and because: drones. In fact, most of 
these thwarted efforts play out in a strikingly similar manner to the earlier clip. Just 11 days 
prior to the 79th releasing the video of the three ill-fated T-72s, they posted this video. And 9 
days prior to that, they posted this one. In each, drones, like Ukraine itself, are punching 
far above their commercial-grade weight. And they’ve been doing so since the very first days 
of the invasion—it just looked a bit different. First it seemed the Turkish-produced Bayraktar TB2 
might carry the day. Capable of staying aloft for up to 27 hours, carrying a payload of up to 330 
pounds or 150 kilograms, these weren’t cheap, as Ukraine purchased six of the drones, 
along with three control stations in 2019

for about $69 million dollars. It was worth 
the investment initially. As early as 2021,   they patrolled the Donbas region and even fired 
on a separatist position. Then, in February,   with the invasion beginning, they quickly reached 
legendary status—successfully firing on tanks,   fuel trains, fighting vehicles, and missile 
systems. The TB2 quickly gained a reputation in   the invasion’s early stages. By April, they were 
now sinking ships while Ukrainian troops sang—and   Ukrainian radio played—the Bayraktar song as 
the Turkish drone had become a central figure of   Ukrainian resistance. The world took notice, too, 
as The New Yorker even went so far as to publish a   story titled “The Turkish Drone That Changed 
the Nature of Warfare.” Then something changed.  Specifically, the global attention 
on the TB2 extended to Russia,   which after anchoring more defense positions 
near and within Ukraine’s borders, directed

more attention to and surface-to-air missiles, 
at the relatively slow, relatively low-altitude, and relatively expensive drones, effectively 
blasting them off the front-lines and into more minor observational roles. 
High-dollar drones worked, but they didn’t play to Ukraine’s advantage—its 
resourcefulness and conviction in the role of the defensive combattant—nor recognized the 
gap in available resources between them and Russia. Fighting with, and inevitably losing such 
expensive equipment played into Russia’s hands. But rather than pivot entirely 
away from UAVs, they iterated, moving away from the military-grade, million 
dollar drones for the unassuming sort; the commercial, cheap, easy-to-operate, and easy 
to produce quadcopters. While DJI, the world’s most renowned commercial quadcopter producer, 
has never made a military-grade drone, and has no interest in its products being used, sold, or 
thought of as weapons, they’ve become exactly that.

in the 21st century’s most significant 
ground war to date. In October of 2023, the country’s prime minister Deny Shmyhal claimed 
that Ukraine had gotten their hands on some 60% of the company’s global output of Mavic quadcopters. 
These drones, DJI or otherwise, play squarely to Ukraine’s strengths. For one, they just don’t cost 
much—they retail at under a thousand dollars. Used as small-area scouts, they also play to the 
advantage of the defender rather than the aggressor, as any advance, build up, or really 
any disturbance on the frontlines becomes easy to monitor via drone while the operator 
maintains their cover. They also help mitigate Ukraine’s ammunition deficiency, as 
a drone can scout targets—while a TB2 can spot a potential target miles away, a quadcopter 
can fly near enough to make sure it is indeed

worth the in-demand artillery shells to attack. 
And this technology has helped Ukraine undercut mighty Russia’s greatest fighting strength: its 
sheer scale. With the advent of such accurate, unrelenting monitoring of every movement, Russian 
forces have had to adjust, moving in smaller numbers more quickly, which, for a fighting force 
known for prevailing by force but consistently plagued by organizational issues, is a big ask. 
And for a fighting force constantly in need of supplies, drones are uniquely easy to crowdsource 
— Ukrainian citizens have been happy to donate their hobby drones to the cause, and so too have 
citizens across the heavily sympathetic West. But if the idea of repurposed quadcopters 
was resourceful on the part of Ukraine, then the advent of mass-produced kamikaze 
drones is nothing short of scrappy. Mechanically, there’s a good few differences 
between reconnaissance quadcopters and kamikaze

drones—some of the latter are fixed 
wing. A vast majority are piloted by fixed camera first-person-viewing systems, and 
increasingly these are manufactured strictly for military purposes within the borders 
of Ukraine. But the biggest difference is that these aren’t capable of carrying, then 
dropping, a payload; they are the payload. On an economic scale, these make obvious sense. 
Consider the earlier example of kamikaze drones ramming into the weak points in downed tanks. 
Now, it’s difficult to boil down the exact unit cost of a T-72; they cost a couple million per 
when built during the Soviet era, and they cost over $200,000 each to ramp up for the standards 
of modern warfare. But whatever the cost, the math remains simple, as the oft-cited going rate 
for a first-person-view kamikaze drone is about $400. Suddenly, the playing fields of an 
asymmetrical conflict becomes a bit closer.

to level. And this goes for human capital, 
too. Ukrainian troops on the eastern end are   outnumbered by orders of magnitude by Russian 
soldiers, so anytime a DJI Mavic can search   the fields surrounding Marinka for retreating 
Russians, or a quadcopter can drop a payload   big enough to finish off a Soviet-era tank, 
Ukraine keeps another soldier out of harm’s way.   Across what’s nearing three years of innovating 
and iterating, drones have become central and   fundamental in Ukraine’s defense. And its military 
knows it. Just take the 79th Air Assault Brigade’s   website: there’s soldiers, there’s a helicopter, 
and there’s a drone. And, should one view the   brigade’s listed vacancies, they’re looking to 
hire more drone pilots at a wage competitive   to the rest of their open positions. The brigade 
has even gone so far as to create an attack drone.

company to flank its more traditional tank company, 
and attack battalions. And along with more pilots, they need more drones, something that 
battalion members have posted on YouTube, and something that American 501c3’s have latched 
on to as an easy way to help the cause, with groups like Ukrainian Defense Support publishing 
explainers on how to get all important drones from American consumer’s hands to Ukrainian soldiers.
But the cycle of military innovation is predictable, and the next stage after a novel 
technology opens up an asymmetrical advantage is the development of countermeasures. In this 
case, some of the countermeasures are stupidly simple. For example: nets. The exposed rotors 
of commercial quadcopters will quickly seize up when in contact with just about anything, 
so simple netting is enough to stop them in their tracks. So facing the new threat, Russia 
has adorned all their key infrastructure near

the front line with so-called anti-drone 
netting, and it’s working. In addition, they’ve experimented with building metal cages 
around high-value vehicles and weapons to at least minimize damage from kamikaze drones—keeping 
the blast further away from fragile components. But then there’s the offensive option. The sorts 
of sub-$1000, commercial drones used in this war have rather limited flight time—between 
20 and 30 minutes—and even more limited signal range—often as little as a mile. While 
there are ways to reduce these limitations, operating kamikaze drones always requires the 
operator to be effectively on the front line. Therefore: drone on drone warfare. Observing 
their effectiveness, Russia has built up an equally-strong drone capability, backed 
by a burgeoning domestic manufacturing industry. Along the front line, operators 
from both sides now hide in makeshift bunkers, peaking out momentarily to launch their aircraft 
on a mission to hunt out their counterparts just

miles away. Finding and destroying an enemy 
drone base is now a prime objective of each side as it has the ability to immobilize a 
whole fleet of potentially destructive drones, rather than just one tank or truck or soldier. 
But perhaps the most effective countermeasure is signal jamming. Cheap commercial drones rely 
on GPS to navigate, but fundamentally what a GPS signal is is a rather weak radio wave broadcast 
from a satellite in space. Therefore, all it takes to disrupt GPS navigation is broadcasting 
a different, incorrect signal on the same frequency. This is what GPS jamming is, and it’s 
now rampant in hotly-contested areas. And the same principles apply for essentially any other form of 
wireless communication. It's all just radio waves of different frequencies, so if Russia knows what 
frequency Ukrainian drones use to communicate with their operator, which is fairly predictable 
if they’re using popular commercial drones,

they can simply overwhelm that frequency with 
irrelevant radio waves, forcing the drone to lose signal and crash. This sort of electromagnetic 
warfare has turned the drone war into a game of cat and mouse. One side develops a signal jammer 
capable of interfering with the frequency used by the other side’s drones, so the other side 
develops drones that communicate using a different frequency, then the first side adapts their 
electronic warfare capabilities, and so on and so on. The net effect is that drones have gotten 
less effective for both sides. The likelihood of a given drone successfully destroying an enemy 
asset has steadily declined, and therefore that incredible efficiency that made headlines in the 
early days of the war is quickly diminishing. But there’s an obvious solution, and it's 
seen in this short clip. These red boxes represent the first days of a new epoch 
of warfare. That’s because this drone,

developed by startup Ukrainian company Saker, is 
autonomously identifying targets. Within each box,   is what a computer vision algorithm believes is 
a target that could be strategically beneficial   to destroy, while the text above indicates what, 
in particular, it thinks it is, and the number to   its right is an indication of the software’s 
confidence in what it believes it sees.   The short-term benefit of autonomy is 
straightforward: Russia’s most effective   countermeasure is to interrupt the signal between 
a drone operator and a drone, so what if the   drone doesn’t need a signal? What if the drone, 
once deployed, could independently navigate to,   identify, and strike a target. Or even: what if it 
could determine its target and decide to strike it   itself without any authorization by an operator? 
While all indications suggest that there’s not yet   wide-scale use of AI drones in Ukraine, Saker’s 
scrappy autonomous drones have reportedly already

destroyed Russian targets in autonomous mode, 
meaning the era of AI warfare has quietly begun. In practice, autonomous drones have yet to make 
a major impact in the war as they still require human involvement. They’re rather finicky, and 
they’re more costly than equally destructive conventional equivalents—but 6,000 miles 
away, on the other side of the Atlantic, in an industrial area next to an Ikea 
in Costa Mesa, California, one company is trying to change that. Its name is Anduril.
Anduril’s heritage explains a lot. Its founder, Palmer Lucky, was the pioneer behind the Oculus 
brand of VR headsets. While still a teenager, he grew this into a burgeoning company and 
eventually sold it to Facebook for $2 billion at just 21-years-old. During these years, others 
that would eventually join Anduril were working at SpaceX and Palantir. The significance of 
this pair of companies is in the fact that

they effectively built the Anduril business model.
That’s because the rocket launch and predictive analytics companies each took the US government
on in court when they believed they were being shut out of competitive bidding for US military
contracts in favor of the old guard of the military industrial complex like ULA or Raytheon.
Each of these companies believed the US military procurement system was broken, and this belief
was well grounded. After all, the United Launch Alliance was paid to keep operating a wildly
inefficient and aged Atlas V launch system for decades, with absolutely no incentive for
innovation in a way that might bring down cost for the government. That’s because, like many military
contracts, ULA was paid on a cost-plus basis, meaning they were paid whatever it cost for them
to do the work they were asked to do, plus a fixed percent for profit. In many ways, this actually
disincentivized innovation since creating a more

efficient system that cost less per launch would actually reduce their fixed profit percentage. But SpaceX wasn’t getting these contracts anyways, so their solution was to foot the cost of innovation themselves, develop a more efficient launch system, then enter a competitive bidding process to offer space access at a lower cost, yet still turn a profit. After some legal tussles, this worked. The government had effectively no choice but to accept their proposal to do the same work for less, and they’ve now grown into the largest launch provider for the US. Anduril was formed under the same model—that of a traditional company, rather than a military contractor. But rather than work on the fringes of the industry, competing in the space-launch or predictive analytics spaces, which have plenty of private customers, Anduril is taking the old-guard head-on—developing innovative products that are generations ahead of what the legacy contractors

are offering, exclusively for the US and allied 
militaries, under the belief that their offerings   will be just too good to pass up. 
At the core of that value-proposition   is artificial intelligence. They seem to 
recognize the shortcomings of early autonomy   in Ukraine—fundamentally, that the full potential 
of autonomous drones is stymied by the persistent   one operator to one device equation. Just as 
vehicle autonomy is still merely a convenience   rather than the promised generation-defining 
breakthrough due to the need for human oversight,   drone autonomy won’t either until it’s able to 
unlock unimaginable degrees of volume. That’s   why Anduril’s marquee product is Lattice—this 
is essentially an operating system… for war.   This promotion video demonstrates how Lattice 
is supposed to work. In this mock scenario,   a combatant drone is detected by the company’s 
Sentry product—one of its first, originally

deployed along the US-Mexico border as part of a 
contract with US Customs and Border Protection. Sentry then alerts an operator, who elects to 
activate Pulsar—Anduril’s electromagnetic warfare solution, capable of jamming communication signals 
to and from the drone. But next we see the launch of Anvil—their kinetic interceptor drone or, put 
another way, the drone built to smash into other drones. Each of these devices work autonomously, 
yet are strung together into an integrated system by Lattice. And Anduril’s has plenty more products 
to add to that system—a jet-powered interceptor, an infrared surveillance platform, and a 
wide variety of other airborne platforms. This is what unlocks the full potential of 
drones. Highly capable drones are now cheap, but human operators are not. So by stringing together 
autonomous drones with an operating system, both the drones and the operation of drones is cheap. 
This is where capabilities really compound.

Destruction in warfare typically follows certain 
rules. A grenade might be cheap and destructive,   but it’s not very capable—it requires 
close proximity. A guided missile might   be destructive and capable, but it’s not very 
cheap—its manufacturing is extraordinarily   expensive. A single kamikaze drone might be cheap 
and capable, but it’s not very destructive—it can   possibly destroy a tank, but with lowering success 
rates, it’s really that a single drone can destroy,   on average, say, a tenth of a tank. 
Interconnected, autonomous drones, however,   are cheap, capable, and massively destructive. 
And that’s largely thanks to drone swarms. Without   the need for operators in close proximity for 
each aircraft, a military could deploy dozens,   hundreds, even thousands of drones without 
a risk to human life on their side before   getting to the cost of a single advanced 
precision-guided missile. That’s to say:

the cost of killing is getting scarily low.
And then there’s one other key difference—to date, essentially every life taken in war has been the direct result of a decision made by another human. Human judgment determines death. But soon, artificial intelligence algorithms might. Humans will decide to deploy a drone, but a drone will be capable of independently determining whether a life is worth taking. So that’s to say, in addition to removing the monetary and human cost of killing, autonomous drones also remove the moral cost—nobody has to bear the weight of pulling the trigger that ends a life. Killing should have friction, it should be costly, it should feel terrible. This new era of warfare unlocks apocalyptic levels of efficiency in death. It is often the case that early observers overestimate the potential calamity that new military innovation will bring—the long-term average is that reality is not as bad as

we fear—but there is a fear that this time might be different. Drone warfare has precedent—we’ve seen how militaries act when they have access to risk-free killing anywhere in the world with multi-million dollar drones manufactured by major contractors. Some of the most horrific actions by the US military have happened outside the context of a formal war through the use of remotely-operated aircraft. Civilian casualties have been enormous, and the state of war is now a blurry, near-perpetual concept—strikes happen indiscriminately in countries with which the US has no active state of war. Now, we’re entering an era where this same technology can be acquired on a miniaturized scale not from military contractors, but from online retailers. So the concern is twofold. First, what will non-state actors—terrorists, cartels, and others with a will to kill—do with a technology that allows them to transport an explosive device effectively

anywhere, at limited risk or cost to themselves. 
And second, with the expanded capabilities of massive swarms of drones, what will state actors 
do when the accountability and friction of war is minimized to perhaps its lowest level ever. 
As this video makes clear, artificial intelligence is becoming quite influential—it is the key 
technology around which the next generation of weapons is being built. When any technology 
becomes this influential, I believe it’s important to have an understanding of how it actually 
works, and the best place to do so is our sponsor, Brilliant.org. That’s because one of their courses 
is called “Introduction to Neural Networks.” This is the exact tool being used in Ukraine right 
now to aid drone pilots in identifying targets, and in this course Brilliant.org helps you go 
from nothing to having a decent understanding of the inner workings of this tech. They do 
so by breaking the subject down into small,

intuitive principles, then teaching these through 
interactive exercises and straightforward visuals. Then, as you move on, they bring these concepts 
together and soon enough, you grasp the overall subject. It really is a better way of learning, 
and they have these excellently designed courses for dozens of STEM-related subjects. So if 
you’re the kind of person who enjoys having an understanding of how things work, Brilliant.org 
is an excellent resource to get a grasp of these subjects that are tough to learn outside the 
context of formal education. They really just make it practical, as you can progress on their courses 
either on the computer or their mobile apps, and a given section only takes 15 minutes or 
so, so you can make progress in understanding, say, gravitational physics, just while waiting 
for the subway. And best of all, you can try everything Brilliant has to offer for free for a 
full 30 days. Just visit Brilliant.org/Wendover.

or click on the link in the description. You’ll also get 20% off an annual premium subscription, and help support this channel, so thanks in advance for checking them out.

The Terrifying Efficiency of Drone Warfare

Wendover Productions is all about explaining how our world works. From travel, to economics, to geography, to marketing and more, every video will leave you with a little better understanding of our world. New videos go out every other Tuesday.

The Terrifying Efficiency of Drone Warfare

The Terrifying Efficiency of Drone Warfare

Scripts