Why scepticism about cellular drones is increasingly difficult to sustain

A view persists in parts of the defence, security and telecommunications community that cellular-controlled drones are a marginal phenomenon: a Ukrainian battlefield curiosity, an interesting Operation Spiderweb footnote, or a theoretical concern overtaken in importance by fibre-optic and mesh-controlled systems. The view is reasonable on its face. It is also, on the available evidence, no longer correct.

In the space of roughly thirty months, cellular-controlled drones have moved from rare experimental adaptations to a recurring feature across an unusually wide spectrum of hostile activity. They have been used for strategic strikes that destroyed elements of a nuclear-armed state's bomber force. They have been integral to long-range one-way attack platforms launched routinely against the cities of a major European country. They have appeared in forensic examinations of recovered front-line drones from a peer-state conflict. They have featured in the criminal economy of prisons in the United Kingdom. And they sit one technical step away from the airspace incursions that have closed major European airports and brought NATO's Article 4 into play.

This article does not argue that every hostile drone is, or will be, cellular-controlled. Many will continue to use direct radio control, mesh, satellite or fibre. It argues something narrower and harder to refute: that across a genuinely diverse set of hostile-use scenarios, by genuinely different actors, with forensic and prosecutorial evidence to support the claim, cellular control is now a recurring tool. Anyone planning around the assumption that it is a niche capability is planning around a threat picture that is at least two years out of date.

What follows is a catalogue of the evidence, organised by the type of hostile activity rather than by the country in which it occurred. The pattern across types is the point.

1. Strategic deep strike: Operation Spiderweb

On 1 June 2025, Ukraine's Security Service launched 117 first-person-view drones from concealed positions inside Russian territory against five strategic air bases distributed across five Russian time zones. The drones were driven into position by unwitting hauliers, and released by remote command. Ukraine claimed approximately forty-one aircraft destroyed or damaged at a cost to Russia of roughly seven billion US dollars; Russia acknowledged eleven losses. The furthest target — Belaya, in Irkutsk — sits some 4,300 kilometres from Ukraine.

The control architecture is the relevant point here. Open-source analysis has converged on a consistent picture: ArduPilot autopilot software running on commodity flight controllers, ESP32-class microcontrollers handling cellular interfacing, communications carried over the Russian 4G/LTE mobile networks operating in the vicinity of each target. The drones did not depend on satellite navigation, which made them immune to GPS jamming. The operators were in Ukraine; the cell sites carrying the control link were Russian civilian infrastructure.

What Operation Spiderweb demonstrated, and what subsequent commentary has only sharpened, is that a cellular-controlled drone strike against high-value strategic assets is not a theoretical possibility. It has been performed, at continental scale, against strategic aviation of a peer state, by a small organisation, using commodity components, and at a financial cost-per-target several orders of magnitude below what the target itself was worth. The technique has been published, photographed, and analysed in the open-source. Replication does not require state intelligence services: it requires off-the-shelf parts and access to whichever public mobile network operates over the target.

2. Long-range one-way attack: the Shahed–Kyivstar pattern

Since late 2023, Iranian-designed Shahed-136 one-way attack drones launched by Russia against Ukrainian cities have routinely been recovered carrying 4G modems and SIM cards procured from the Ukrainian mobile operator Kyivstar. The pattern was first publicly confirmed by Ukrainian Air Force spokesperson Yuriy Ignat in late 2023 and has been corroborated by Ukrainian drone-defence specialists and Western analysts including Jack Watling of the Royal United Services Institute.

The exact operational purpose of these SIMs is still partly debated. The most plausible uses combine several functions: cellular-tower triangulation as a navigation backstop against Ukrainian GPS jamming; real-time route adjustment, allowing operators to update targeting after launch; post-strike confirmation that a drone has reached its destination; and, in some cases, limited signals-intelligence collection. Specific cases have also surfaced of Ukrainian counter-intelligence arresting individuals supplying Ukrainian SIM cards to Russia for this purpose, most recently in September 2025.

The category here matters. Unlike Spiderweb, which used cellular as primary control, Shahed-type drones use cellular as resilience layer and augmentation. This is the pattern most likely to spread to other long-range attack platforms operated by other actors: the SIM does not need to do everything; it needs to do enough to give the platform an additional capability the defender must contend with.

3. Front-line and mid-tier military innovation: the Russian "Orbita" and mothership patterns

Forensic examination of recovered drone wreckage in Ukraine through 2025 has confirmed Russian use of LTE-controlled FPV systems, reportedly under the project name "Orbita." Photographs of exposed electronics including SIM cards and 4G/LTE modem modules in FPV airframes were published on Telegram by Serhii Beskrestnov, an adviser to the Ukrainian Defence Minister, in October 2025. The technique was used in combination with so-called "mothership" drones — larger platforms that carry smaller FPVs forward and release them closer to the target, extending effective FPV range from the typical 20–30 kilometres to substantially further.

Ukrainian authorities have responded by considering passport-based registration for SIM card sales, on the basis that anonymous SIMs are themselves now a force-protection vulnerability. The technique has spread sufficiently that, in February 2026, the Kyiv government was openly weighing changes to commercial SIM regulation in direct response.

The wider implication is that even in active high-attrition front-line warfare — the place where mobile networks are most likely to be jammed, destroyed or denied — cellular control is being used wherever coverage survives. It is not a substitute for radio or mesh; it is an addition to them. Operators choose whichever channel works in a given engagement.

4. Hybrid grey-zone incursions: Europe, 2025

Across September and October 2025, Europe experienced an unprecedented sequence of hostile drone incidents against critical infrastructure and civil aviation.

On 9 September, approximately twenty drones penetrated Polish airspace, closing four airports including Warsaw Chopin and triggering the first invocation of NATO's Article 4 since the start of Russia's full-scale invasion of Ukraine. Within days, drone activity forced closures at Copenhagen and Oslo airports. In late September, German authorities reported overflights of the Kiel naval base, the Heide oil refinery, hospitals, and shipyards. Through October and into November, suspected hostile drones were observed in Denmark, Norway, Estonia, Latvia, Lithuania, Romania and Sweden. The European Parliament adopted a formal resolution on 9 October 2025, by 469 votes to 97, condemning the pattern of incursions and identifying it as systematic hybrid warfare. Between September 2025 and January 2026, one analytic group recorded thirty suspected or confirmed NATO airspace violations — more than the total of the preceding three and a half years combined.

The cellular dimension of these incidents is, currently, partly an inference rather than a fully forensic finding. Many of the platforms involved were larger, longer-range systems with their own dedicated communications. But several characteristics of the pattern — small drones operating in civilian airspace, launches from inside the targeted countries themselves, persistent difficulty in attribution despite extensive investigation, and the use of vessels of the Russian "shadow fleet" as suspected launch platforms — are exactly the operational characteristics that cellular control enables. A drone with a local SIM card, launched from inside the target country by an operator who flies home the following morning, leaves none of the diplomatic traces of a state-launched weapon crossing a border. Plausible deniability is genuine. Attribution is genuinely difficult.

The Russian state's own response in late 2025 is the most telling indicator of where the threat is understood to lie. Confronted with sustained Ukrainian drone activity against rear-area infrastructure, Russia began imposing rolling mobile internet shutdowns from May 2025. By November, fifty-seven Russian regions were reporting daily cellular disruptions, in many cases lasting days at a time. Foreign SIMs entering the country are subjected to a 24-hour data-and-messaging cooling-off period; returning Russian citizens face the same restriction on domestic SIMs after international travel. The justification offered openly by Russian authorities is that hostile drones use the cellular network for navigation and control, and that denying the network denies the drones. The civilian cost of these measures — ATMs failing, pharmacy software broken, public transport ticketing collapsing, glucose monitoring lost in homes with diabetic children — has been substantial. A state does not impose costs of that magnitude on its own digital economy in response to an imaginary threat.

5. Organised crime: the United Kingdom prison economy

In the year to March 2025, His Majesty's Prison and Probation Service recorded 1,712 drone incidents at prisons in England and Wales — a 43 per cent increase on the previous year and an approximately 770 per cent increase on 2019. HMPPS has publicly stated that an increasing proportion of these drones are cellular-equipped, allowing operators to fly from miles away rather than from a vantage point overlooking the prison wall. In June 2026, the UK government announced £35 million in new counter-drone investment for the prison estate, on top of the existing £40 million security budget.

The Chief Inspector of Prisons has publicly stated that "the prison service has, in effect, ceded the airspace above many of our prisons to organised crime gangs."

The point for sceptics is straightforward. The same technical capability used by a sovereign state intelligence service in a continental strategic strike — a small drone, a cellular modem, a SIM card, a remote operator — is being used by London-based criminal gangs to run an industrial-scale contraband economy into British prisons. The hostile actor differs by orders of magnitude. The underlying technology does not.

6. The wider criminal and border picture

In Mexico, weaponised drone use by cartels has grown sharply. Between 2021 and 2025, the National Counterterrorism Innovation, Technology and Education Center (NCITE) identified 221 cartel drone incidents, of which 27 resulted in fatalities. The Jalisco New Generation Cartel (CJNG) accounts for approximately one in five of these. In October 2025, three explosive-laden drones struck the offices of a state prosecutor in Baja California. US Customs and Border Protection has reported tens of thousands of drone sightings at the southwest US border, with cartels using drones for smuggling, surveillance, and direct attacks on Mexican security forces.

Forensic evidence of cellular control specifically in the Mexican cartel context remains thinner than in Ukraine. Most documented cartel drones to date have used standard radio control or commercial autopilot systems. But the operational logic that pushed Russian and Ukrainian forces towards cellular control — extending range, evading jamming, allowing operators to remain at a safe distance — applies identically to cross-border drug smuggling and to attacks against fortified police and military positions. The commercial enabling kit is already in cartel hands; CBP regularly reports recovery of drones with extensive aftermarket modification. The probability that cellular control is not, by some point in the next two to three years, a routine feature of cartel drone operations is low.

A similar pattern exists in other contested regions — the Sahel, the Sudanese civil war, the India–Pakistan border, the Israel-Gaza-Lebanon theatre — where drone use is well established and where the broader adaptation curve from RC to cellular has yet to be publicly documented but is operationally plausible.

7. The commodity enabler

A critical point for sceptics evaluating future risk is that the technical floor for entering cellular-controlled drone operations is now extremely low.

Off-the-shelf commercial products allow it directly. DJI's Cellular Dongle 2, retailing at approximately $150, provides a sanctioned 4G connection for the company's enterprise drones — a product marketed for legitimate inspection, delivery and emergency-response use cases but operationally available to anyone who buys one. Beyond the commercial path, hostile assemblies seen in Ukrainian forensic analysis typically consist of an ESP32-class microcontroller, a generic LTE modem module, an antenna and a locally-purchased SIM card. Total component cost is in the tens of dollars. Software is open-source: ArduPilot, the autopilot used in the Spiderweb operation, is free.

The standards body 3GPP has, through Releases 15 to 18, progressively added explicit support for aerial vehicles to the 4G and 5G specifications, on the basis that legitimate drone delivery, agricultural monitoring and infrastructure inspection use cases require it. The same standards work that legitimises civilian drone integration onto mobile networks lowers the technical threshold for hostile use. There is no licensing, no special hardware, no exotic spectrum allocation. A modern cellular-connected drone is, in materials and effort, indistinguishable from a smartphone with an electric motor strapped to it.

Why cellular drones generalise across scenarios

The case-by-case evidence above is the centre of the argument. The structural reasons it generalises across so many different scenarios are worth stating explicitly.

Cellular drones offer three properties no other control modality combines.

First, range. A drone on a public mobile network can be flown wherever cellular coverage exists. In most developed economies that means the entire national territory and much of the surrounding region. The operator can be in a different city, a different country, or a different continent. Direct radio control reaches kilometres; cellular reaches countries.

Second, jam-resistance against conventional countermeasures. Most counter-drone electronic warfare developed in the past decade has targeted the 2.4 GHz and 5.8 GHz bands used by commercial drone remote controls. Cellular drones do not use those bands. Jamming a national mobile network sufficiently to disrupt them means denying service to the civilian economy that depends on it — a step democracies cannot take routinely and authoritarian states can take only at substantial cost. The Russian internet shutdowns of late 2025 represent the upper bound of how far a state will go, and the civilian damage they have caused suggests the upper bound is being tested.

Third, deniability. A cellular-connected drone leaves a forensic trail in the network, but that trail accrues to the network operator rather than to the launching state. A drone launched from inside the target country, using local SIMs, carrying no markings, does not provide the diplomatic clarity of a missile or aircraft crossing a border. For an actor seeking to operate below the threshold of declared war, this is precisely the quality being sought.

These properties combine across attacker types. A state intelligence service wants strategic depth and post-strike deniability. A long-range strike force wants jam-resistance and geolocation backup. An organised crime group wants range from a safe operator location. A hybrid-warfare planner wants attributability gaps. The same architecture serves all of them.

Addressing the front-line objection

One sceptical objection deserves to be addressed directly: that cellular networks do not survive contested front-line conditions, and that cellular-controlled drones are therefore irrelevant where serious fighting actually occurs.

This objection has the geography of the problem backwards.

In a country at war, the contested front line is a relatively small fraction of the total territory. The rear areas — the strategic depth, the cities, the airbases, the logistics nodes, the political centres, the civilian infrastructure — are vastly larger and remain on civilian mobile networks for the duration of the war. Operation Spiderweb attacked targets up to 4,300 kilometres behind the formal front line. Russian Shahed strikes on Kyiv use Ukrainian mobile networks not at the trench line but inside cities. The hybrid drone incidents at Copenhagen, Oslo, Warsaw and Kiel did not take place in a contested zone at all.

Where cellular networks are intact, cellular drones are usable. Where they are not, drones use other control channels. The choice is not "cellular or nothing"; it is "cellular where it works, alongside everything else where it doesn't." The front-line objection identifies a real operational constraint and then misreads it as a strategic dismissal. The places where cellular drones cannot operate are precisely the places that matter least for strategic effect. The places where they can operate — civilian territory, strategic infrastructure, dense urban areas, ports, airports — are precisely the high-value targets the attacker most wants to reach.

A second variant of the objection notes that wartime conditions can prompt states to shut down their own networks, as Russia has done. This is true and worth examining. The Russian shutdowns of 2025 are the most extensive state-directed cellular degradation imposed in peacetime in modern history. They have not stopped Ukrainian drone activity, only complicated it. They have imposed enormous costs on the Russian civilian population. For a democracy, the same tool is not available at remotely comparable scale. The defender's problem is therefore not whether to shut the network down — it is how to find hostile drones operating on a network that must remain open.

Trajectory

The trajectory of cellular drone use is the part of the picture that most undermines a position of scepticism. Across the indicators that determine whether a technique remains marginal or becomes mainstream, all of them point the same way.

The hardware barrier is falling. Off-the-shelf commercial cellular dongles are widely available. Bespoke modules can be sourced for tens of dollars per drone. The cost differential against radio-controlled drones is negligible. The capability uplift is substantial.

The actor pool is broadening. The technique began with state intelligence services and front-line military innovation. It has spread to organised crime in advanced economies, to non-state armed groups in active conflicts, and into the toolkit of hybrid-warfare planners targeting countries that are not at war. There is no plausible reason to expect this widening to reverse.

The operational success record is substantial. Operation Spiderweb has been described in serious analyses as the most cost-effective strategic strike in modern military history. The Shahed-SIM pattern continues. The grey-zone incursions in Europe have so far been politically consequential without prompting effective response. Every demonstration of effect lowers the barrier for the next actor to attempt the technique.

The standards trajectory is supportive of hostile use. 3GPP's support for cellular drones on civilian networks is mature and improving. The same trajectory that legitimises drone delivery for hospitals, farms and logistics companies provides the standards backbone for hostile use of the same networks. Civilian and hostile uses share the same infrastructure.

The defensive arrangements have not yet caught up. Most national counter-drone postures remain organised around radar, RF sensing in the consumer bands, and physical intercept. These are necessary but they were designed for a class of threat that emits in known military bands, is large enough to radar-paint, and is flown from within line of sight. A cellular-controlled drone breaks each of those assumptions. Closing the gap requires bringing telecommunications operators, regulators and security stakeholders into a structured working relationship that, in most jurisdictions, has not yet been formed.

Conclusion

The most useful single observation about cellular-controlled drones is that they are now documented across an unusual range of hostile-use scenarios — strategic strike, long-range attack, front-line military innovation, hybrid airspace incursion, organised criminal economy, and emerging cross-border smuggling — by actors that share no political alignment, no operational doctrine and no common procurement chain. What they share is access to the same civilian infrastructure and the same commodity hardware. That shared substrate, more than any of the individual examples, is the evidence that should shift the burden of proof.

Scepticism about cellular drones was a reasonable position in 2023. It is increasingly difficult to sustain in 2026. The pattern is documented, the forensics are public, the actor base is widening, the commercial enabling kit is mainstream, and the strategic effects have been demonstrated at continental scale. The defensive challenge is not to debate whether the threat is real — that argument is, for practical purposes, now settled. It is to decide what to do about a class of hostile drone that is intrinsically national in scale, jam-resistant against conventional countermeasures, deniably attributable, and arriving over the same network that carries the rest of civilian life.

The networks are not going away. The drones are not going away. The serious question is whether those responsible for national resilience will look at the signalling that already exists on the civilian networks the attacker depends upon — or wait for a Spiderweb-class event to take place somewhere closer to home, and decide then.