Over four years small drones in Ukraine have evolved over from novelties into essential assets. They now dominate the battlefield, ubiquitous eyes in the sky which also inflict around 80% of all casualties. Recent advances have made drones more autonomous, and the way seems open for the next big step forward: swarming, breaking the one-operator-per-drone limit to control an entire swarm at once.

Latvian-Ukrainian defense technology company DK NEJET recently set out to establish the way forward for their drone swarm technology . The challenge is not what you think.

“Algorithmically, a lot of swarming is solved,” Dr. Jan Dyre Bjerknes , Founder and Principal of Framsyni based in Norway, told me. “The challenge is to find the right operational concepts to make it trustworthy and reliable for the users.”

DK NEJET held an event with military users in Ukraine to discuss swarm drones, supported by Dr Bjerknes, an academic in the field of swarming robotics with 16 years’ experience in the defence industry. The company’s FPV drones are already used in Ukraine, but swarming is something else.

A swarm is not just a collection of drones, but a collaborative, co-operative group able to work together. It requires a high degree of autonomy in the drones, and the room quickly divided between those who were enthusiastic about the potential of swarming and those who rejected the idea completely.

“We have had some forceful pushback, people got very engaged,” says Dr Bjerknes. “Some of them have had bad experiences with drone autonomy and they do not want to hear about it ever again.”

In 2024, there was considerable hype surrounding AI-enabled FPV drones which could supposedly be locked on to a target and could track it, home in, plot a route and score a hit every time. Optimistic developers believed that their systems were ready for the battlefield, but the limitations of simple, cameras, cheap processors, and hastily-modified commercial software became painfully obvious .

In 2026 Ukraine started deploying a new generation of far more capable AI-enabled drones . But there is still considerable skepticism over autonomous drones. Much of the resistance comes from FPV operators who do not want to be displaced by machines which will not do the job as well as humans.

“Drone operators are proud of their skills and they push back against an approach where they feel like skills no longer matter,” says DK NEJET’s marketing manager, who goes by Zee.

Developers of autonomous, swarming drones like DK NEJET will not be able to get products into the field unless they can work with military users and win their confidence, establish how the swarms will be used operationally and prove their reliability.

“The swarming technology itself is only about 20-30%,” says Dr Bjerknes. “The rest is in building the concept of operations, training, and the supporting technology.”

The Missing Piece: Concept Of Operations

Developing a concept of operations, CONOPS in military jargon, is an essential step in introducing any new weapon. It is a detailed blueprint for how a new piece of hardware will be used, how the unit using it is organized, who controls it, and how it fits into the larger scheme.

For example, the Switchblade 300 and Stinger missile are both tube-launched, shoulder-fired weapons used by U.S. foot soldiers. They may look similar but have very different CONOPS.

The Switchblade is part of a rifle company’s assets, and may be used by a line infantryman under the direct control of a platoon commander. Stinger operators are air defence specialists, belonging to Short-Range Air Defence (SHORAD) Battalions. A brigade commander will assign a Stinger team to support a particular unit, but they will still be under the control of an Air Defense Airspace Management cell.

The concept of operations can transform a weapon’s effectiveness. In WW1, tanks were deployed in small groups to support infantry, advancing towards trenches at walking pace. They were there to flatten barbed wire and fire on enemy positions that threatened the foot soldiers. By WWII tank doctrine had transformed: tanks were concentrated into armored divisions, breaking through enemy lines at high speed, making rapid, sweeping maneuvers in a new Blitzkrieg warfare, with infantry following in trucks or half-tracks.

Swarming drones raise as many questions about CONOPS as tanks. One basic issue is over centralization—should they all be under the control of Uncrewed Systems Forces, or individual army units? Should they be organized into different echelons for close support, middle and deep strike? Should separate swarm units be tasked with suppressing air defence and artillery, or will these be the same units that target armor and infantry and attack fortifications? Will interceptor and dogfighting drones be integrated with the swarm, or part of a separate air-defense command?

FPV operators are based near the front line and work closely with ground troops. But they are high-value targets under constant threat from enemy drones and artillery. New technology means the operators can be based hundreds of kilometers away; they would be safe, but will they be able to coordinate as effectively with other units? Or would having a pool of highly skilled operators able to take over drones anywhere in the theater of operations outweigh the disadvantages?

An effective concept of operations is the missing piece in drone swarm deployment.

As Dr Bjerknes notes, the way the swarm works will affect the concept of operations, so the two need to be designed together from the start. Early systems which enable one operator to control multiple drones, such as the Nemyx system developed by Auterion (demonstrated, but not yet deployed in Ukraine), tend to follow a particular pattern.

“Several of the existing developments seem to be variations of the same sort of concept: leader-follower, where the pilot controls one drone and other drones follow, or the system flies drones into an area and then hands over control to a pilot,” says Dr Bjerknes.

In this scheme, attack drones are expendable. Another concept would have unused attack drones return for reuse, but again this has organizational implications. This may become particularly important where targets are dispersed and a swarm of a thousand drones may not find a thousand targets.

In the DK NEJET Swarm, drones pass data between each other and make collective decisions over what action to take. Again, a concept of operations is needed to determine how this should work. For example, if the operator identifies a target and orders a strike, which drone carries it out – the nearest drone, or the one with the lowest remaining battery charge? This type of design decision needs involvement and buy-in from the military users.

The DK NEJET Swarm should be ready for deployment in six months. Initially the company is working with swarms of 32 drones and one operator. Whether their swarms are deployed depends on winning over skeptics and collaborating on a concept of operations. After a slow start, progress is gaining pace.

“We have had tests of our MVP swarm since summer, now in October we see that military people are starting to be more interested in practical swarming and talk about specific use cases,“ says Zee.

“We recently became unexpectedly popular in Russia when they found out about our swarm technology,” says Zee. “From what I’ve seen they have not got anything like that.”

While Russian developers have announced various drone-swarm initiatives, none appears close to deployment. But previous experience has shown that development in small drone technology by one side, from fiber optic control to dragon drones, can be picked up by the other.

“Every time we have a tech advantage they copy it quickly. and it’s just a matter of time,” says Zee.

The Russians may be able to copy the hardware and even the swarming algorithms. But their centralized, bureaucratic system may struggle with the challenge of incorporating direct feedback and criticism from field commanders, something which is essential to rapid, efficient drone development.

And while DK NEJET and the Ukrainian drone operators they talked to are insistent that human operators must remain very much in control, the Russians may have different ideas. Russia has already deployed autonomous hunter-killer drones with very limited ability to discriminate targets. This may make any Russian swarms more dangerous rather than more successful.

An effective concept of operations will determine whether drone swarms become just another system bolted onto existing military structures and tactics, like the tanks in WW1, or whether they will prove as transformative as tanks became in WWII.