NATO’s Theory of Victory: From Ambition to Warfighting Reality. PART 1- A Theory of Victory Is Not Yet a Capability

Eva Sula, Defence & Security Leader | Strategic Advisor | NATO & EU Innovation | True deterrence is collaboration (Tallinn, Harjumaa, Estonia)

1. The proposal: what NATO is actually trying to do

On 5 October, Bloomberg reported that NATO’s Supreme Allied Commander Europe, General Alexus Grynkewich, is preparing what is being called a “Theory of Victory” for defeating a Russian attack on NATO’s eastern flank. An unclassified version reviewed by Bloomberg is expected to be presented in Poland on 12 October. Public reporting describes something more substantial than the headline shorthand of defeating Russia “with drones and AI”: an attempt to explain how NATO would combine existing military power with technologies that have developed rapidly during the war in Ukraine.

The reported sequence matters. NATO would first seek to stop an incursion with land forces and capabilities developed through the Eastern Flank Deterrence Initiative. The latter is intended to counter an adversary’s mass and momentum using scalable technologies including low-cost expendable uncrewed systems, AI-supported target detection and layered defences. Having gained sufficient freedom of manoeuvre, NATO would then seek to disrupt or destroy the adversary’s command, control, communications, computers, cyber, intelligence, surveillance, reconnaissance and targeting system, or C5ISRT, using kinetic and digital means. Conventional air and missile-delivered firepower remains part of defeating the wider Russian combat system.

That is an important distinction. NATO is not proposing to replace tanks, aircraft, artillery, missiles, soldiers or sailors with swarms of autonomous machines. Nor is it starting from an empty military architecture. NATO already has a family of regional defence plans, forward forces, reinforcement arrangements and a capability-planning process intended to produce the forces required for collective defence. Its deterrence and defence posture deliberately combines conventional, nuclear and missile-defence capabilities with space and cyber capabilities. Most of the systems from which those capabilities are assembled remain nationally owned, procured and operated, which will become important when we examine what integration actually requires.

The attraction of adding inexpensive mass, uncrewed systems, automation and AI-supported processing to that architecture is not difficult to understand. Ukraine has demonstrated how extensively relatively inexpensive drones can contribute to surveillance, targeting, strike and attrition. Russia has simultaneously demonstrated that it can produce and employ uncrewed systems at scale and continuously modify them in response to Ukrainian countermeasures. Its Geran programme, for example, has evolved through changes to navigation, communications, propulsion, payloads and tactics, with battlefield feedback feeding further production changes. The competitive advantage is therefore increasingly found not only in possessing a particular system, but in being able to modify and integrate it faster than the opponent can counter it.

Ukraine has also provided a warning against drawing the opposite conclusion. Heavy reliance on drones has partly reflected shortages of personnel, ammunition and conventional equipment, rather than evidence that those capabilities have become obsolete. Ukrainian commanders continue to value artillery, anti-tank weapons, air defence and other conventional effects where they are available, while recent battlefield analysis points towards better results when drones are integrated with infantry, artillery and other tools rather than treated as substitutes for them. NATO therefore has little reason to choose between traditional military power and emerging technology. Its problem is making them work together.

This is where the headline becomes less useful. A drone is a platform. An AI model is a tool. A radar is a sensor. None independently constitutes the military capability described in a theory of victory. The relevant capability is the system that can find an adversary, establish what has been found with sufficient confidence, move usable information to whoever needs it, make a lawful and operationally sound decision, assign an appropriate effect, execute it, determine what happened and adapt when the adversary changes its behaviour. That system also has to continue functioning while the adversary is deliberately attacking its communications, sensors, software, positioning, logistics, infrastructure and people.

Before asking whether NATO has enough drones or sufficiently advanced AI, then, several more fundamental questions need answers.

What does victory mean? Stopping Russian forces at the border, ejecting forces that have entered Allied territory, destroying the attacking formation and compelling Moscow to terminate the conflict are different military outcomes. None automatically produces a sustainable political settlement. A theory of victory needs a sufficiently precise political objective because force requirements, escalation decisions, targeting priorities and the conditions for terminating military operations follow from it.

How quickly is “quickly”? A campaign expected to last days creates different assumptions about ammunition, repair, reinforcement, personnel rotation, industrial replacement and political endurance from one lasting months or years. Ukraine should have made us particularly cautious about treating duration as a secondary planning variable.

Against which Russian course of action is the theory optimised? A large conventional attack presents one problem. A limited incursion intended to test Alliance cohesion presents another. Hybrid activity combined with military coercion presents another still. Russia does not have to select the scenario NATO finds easiest to recognise or answer.

Where does battlefield success become war termination? Destroying formations and command systems can produce military advantage, but an adversary retains political agency. The question is not merely how NATO wins engagements. It is how military success creates conditions under which the adversary concludes that continuing the war offers less than ending it.

These questions do not invalidate the reported Theory of Victory. They are necessary precisely because it deserves to be treated as a warfighting proposition rather than a technology story. Grynkewich’s spokesperson described its underlying purpose simply: maintaining NATO’s advantage and setting the conditions to deter aggression and win if attacked.

The harder work begins when we ask what those conditions actually are.

2. Why 2022 changed the calculation, but did not abolish warfare

Russia’s full-scale invasion of Ukraine did not suddenly make drones, electronic warfare, commercial technology or battlefield software important. All existed before February 2022. What the war changed was their scale, density and interaction. Technologies previously concentrated in specialist units or expensive platforms have spread much further down military organisations, while the time between observing a battlefield problem, developing a countermeasure and encountering the next adaptation has compressed sharply. NATO itself now describes Ukraine as a source of lessons on how inexpensive drones can help offset conventional disadvantages, while its NATO-Ukraine Joint Analysis, Training and Education Centre was established partly to transfer those wartime lessons into Allied training and capability development.

Cheap mass and cost asymmetry. Ukraine planned to supply its armed forces with three million FPV drones during 2025 alone, almost two and a half times the previous year’s number. Scale at that level changes the economics of surveillance and attack. A relatively inexpensive expendable system can threaten vehicles, artillery positions, communications equipment or personnel whose replacement cost is many times higher. It can also force an opponent to spend scarce interceptors or other countermeasures simply to remain protected. The limitation is equally important: cheap individual platforms do not produce cheap warfare. Millions of drones require batteries, explosives, components, operators, communications, repair, transport, software and continuous replacement. Industrial scalability therefore becomes part of operational capability rather than a procurement problem left behind the front.

Persistent, distributed ISR and the growing importance of signatures. Small unmanned aircraft have made observation far more persistent and geographically distributed. Forces can now maintain surveillance over areas that previously required scarce specialist assets, while drones can support target acquisition and battle-damage assessment at tactical level. The operational consequence is a battlefield increasingly hostile to concentration, predictable movement and detectable signatures. CSIS notes that cheap ISR drones have made large visible formations more vulnerable, while RUSI’s work on current combined-arms operations describes Russian forces deliberately using UAVs to locate Ukrainian EW positions, radar, command posts, drone crews and artillery before attacking them with a mixture of fires. Concealment now concerns electromagnetic and digital signatures as much as camouflage.

FPVs, one-way attack systems and shortened engagement cycles. FPVs have made precision attack available at a scale previously impossible for many tactical units. They can pursue moving targets and engage equipment that might not justify expenditure of a more expensive guided weapon. Software and increasingly automated processing can further compress the time between detection and engagement; CSIS has documented Ukrainian cases where AI-supported processes reduced detection-to-destruction timelines to little more than 30 seconds. But FPVs remain constrained by weather, payload, range, communications and EW. Their pilots themselves create signatures and can be targeted. RUSI’s battlefield research also finds circumstances in which heavy EW makes conventional artillery disproportionately useful precisely because the electromagnetic environment interferes with friendly drones as well as hostile ones.

Autonomy, AI and software-driven adaptation. The important development is not a sudden arrival of fully autonomous warfare. Ukraine has fielded narrower functions including automated take-off and landing, computer-assisted target recognition and navigation intended to reduce dependence on vulnerable communications. Russia is moving in a similar direction under pressure from GPS denial, jamming and attrition, increasingly embedding task-specific machine-learning functions at the tactical edge. Software can therefore change the behaviour of an existing physical platform without waiting for a new generation of hardware. That creates obvious advantages, but also a new sustainment burden: software versions, training data, cybersecurity, compatibility, validation and configuration become operational concerns. A drone updated yesterday may no longer behave exactly like the nominally identical system tested three months earlier.

Electronic warfare and counter-adaptation. Ukraine has repeatedly demonstrated why no technological advantage should be treated as permanent. Radio-controlled drones encouraged increasingly dense jamming. Jamming encouraged frequency changes, alternative navigation, greater autonomy and fibre-optic control. Those adaptations produced new detection and protection requirements. Russia and Ukraine are therefore engaged in a continuous interaction between measure and countermeasure rather than progressing towards a final technological solution. Leaders consequently need organisations capable not merely of acquiring innovation but of repeatedly testing, modifying, distributing and retraining around it. CSIS noted as early as 2024 that updates to Ukrainian unmanned systems could emerge within days; by 2026, the institutional question is increasingly how those cycles can be sustained at military scale.

Commercial technology has altered the route from idea to battlefield. Ukraine has made extensive use of commercial components, civilian engineering, open-source technologies and procurement mechanisms that allow operational units to acquire equipment far faster than traditional military acquisition cycles. CSIS reports that Ukraine reduced some unmanned-system procurement timelines from months or years to weeks, while commercial technology now represents a substantial share of Ukrainian defence procurement. The advantage is speed and access to a much larger innovation base. The corresponding risks include supply-chain dependence, component provenance, cybersecurity, inconsistent configuration, intellectual-property dependencies and systems that were never designed for prolonged operation under deliberate military attack.

None of this demonstrates that artillery, armour, infantry, engineering, air defence, logistics or electronic warfare have become obsolete. Current Ukrainian experience increasingly points in the opposite direction. RUSI’s 2025 battlefield research found that bad weather and thermal crossover can degrade drone effectiveness while conventional artillery remains effective against preregistered positions, and that artillery’s larger payload remains important against hardened targets. By June 2026, RUSI’s Jack Watling assessed improved Ukrainian battlefield performance as stemming less from drones alone than from better integration of drones with infantry, artillery and other capabilities. The lesson is therefore not replacement but recombination: established military functions increasingly have to operate inside a battlespace saturated with sensors, software, unmanned systems and rapidly changing countermeasures.

That distinction also makes “combat-proven in Ukraine” a much less useful description than it first appears. A system may have successfully operated with one Ukrainian formation, on one sector of the front, against a particular EW configuration, using a particular software build and supported by experienced operators. That tells us something valuable, but it does not tell us whether the same system remains effective after the adversary adapts, whether another unit can reproduce the result, whether thousands can be supported simultaneously, or whether it can exchange data and tasking with NATO systems during an Article 5 operation.

For the purposes of this series, we therefore need a stricter vocabulary. Demonstrated means something has been shown to work under specified conditions. Tested means its performance has been deliberately examined against defined criteria. Validated requires evidence that it performs the intended function under conditions relevant to its intended use. Fielded means forces actually possess and operate it. Integrated means it functions within the wider operational system on which its usefulness depends. Scalable asks whether the effect can be reproduced at the numbers and organisational level required. Sustainable asks whether personnel, components, software, ammunition, maintenance, training and industrial capacity can keep it functioning through the duration of the operation.

Ukraine has given NATO an extraordinary body of evidence about modern warfare. It has also demonstrated why evidence needs context. The question is no longer whether a drone, algorithm or software application can work somewhere on a battlefield. For a theory of victory, the relevant question is whether the entire chain can work repeatedly, at Allied scale, under conditions chosen partly by an adversary that is learning at the same time.

3. A drone is not a capability

Calling a drone a capability is convenient procurement language, but operationally it is incomplete. The same applies to an AI model, radar, satellite, artillery system or command-and-control application. Each can contribute to a military capability, sometimes decisively, but military effect is produced by the chain connecting observation to action and action to subsequent adaptation. NATO’s own Digital Transformation Implementation Strategy describes its future Digital Backbone in similar terms: a federated architecture intended to connect sensors, decision-makers, actors and effectors across national, organisational, operational and security boundaries. The difficult word in that description is not digital. It is connect.

For this series, the chain can be expressed as Sense → Identify → Understand → Decide → Task → Effect → Assess → Adapt. It is deliberately broader than a conventional kill chain because not every military effect is kinetic, not every detected object should become a target, and the process does not finish when a weapon is released.

Sense begins with detecting something relevant. That may come from radar, electro-optical or infrared sensors, acoustic systems, electronic support measures, satellites, uncrewed platforms, human reporting or several sources simultaneously. More sensors can improve coverage, but they also produce more data, conflicting observations and opportunities for deception. Sensors have different ranges, resolutions, refresh rates and vulnerabilities. Weather affects some. Terrain masks others. Electronic attack can degrade or manipulate them. The operational requirement is therefore not simply to collect more information, but to establish which observations remain sufficiently reliable in the environment in which forces are actually fighting.

Identify turns detection into something on which further action can responsibly be based. A radar return or image may establish that something exists without establishing what it is, whose it is, what it is doing or whether it presents a threat. NATO’s 2026 Counter-UAS Data Challenge illustrates how unresolved this problem remains: NCIA is specifically seeking approaches that combine radar, radio-frequency, camera and acoustic data to improve detection, tracking, classification and identification of small drones. The challenge exists because individual sensors can be ambiguous and because classification errors have operational consequences.

Understand adds context. Knowing that an object has been classified as a particular aircraft, vehicle or emitter does not establish its intent, operational significance or relationship to everything else occurring around it. Commanders need to understand whether a track is hostile, friendly, civilian, deceptive, duplicated or simply irrelevant to the mission. They also need confidence in where the information came from, how current it is and what uncertainty remains. AI can help process quantities of information that humans cannot examine individually, but faster classification does not remove the requirement for context. It can simply allow a wrong conclusion to travel through the system faster.

Decide introduces authority and judgement. Someone or something must determine what action, if any, is appropriate. The technically available response may not be tactically sensible, legally permissible or politically authorised. Rules of engagement, delegated authorities, collateral-damage considerations, national caveats and escalation risk all enter here. Speed matters, particularly against fast-moving or numerous threats, but reducing decision time is useful only if the decision remains sufficiently informed and accountable. A human clicking “approve” on machine recommendations faster than meaningful scrutiny is possible does not necessarily constitute meaningful human control.

Task is where elegant architecture diagrams often encounter institutional reality. Once a decision has been made, the appropriate unit or system has to receive an executable instruction. That can require information to cross service, national, classification and technical boundaries. NATO’s current digital strategy explicitly identifies interoperability, federated identity and access management, standardised data labelling, machine-readable standards and the rationalisation of legacy networks as necessary components of digital transformation. If the sensor can detect a target but the resulting data cannot enter the effector’s system, or cannot legally be released to the nation operating it, the chain stops despite every individual platform functioning correctly.

Effect asks a more demanding question than whether something was launched. Did the action produce the required military result? An effect may be kinetic, electronic, cyber, informational or protective, and the appropriate choice depends on the target and mission. Destroying an inexpensive drone with a much more expensive interceptor can be tactically successful while creating an unsustainable exchange ratio. Jamming may suppress a threat without destroying it but can also interfere with friendly systems. An exquisite weapon may achieve precisely the required effect but exist in quantities too small for sustained use. Operational effectiveness therefore includes cost, availability, opportunity cost and what must remain available for the next engagement.

Assess determines what actually happened. Battle-damage assessment has always been difficult, and ubiquitous sensors do not eliminate the problem. A video feed showing an impact does not necessarily demonstrate destruction, while disrupted communications may prevent the force that generated an effect from receiving subsequent sensor information. Assessment also feeds future targeting and resource allocation. Without reliable assessment, forces can repeatedly engage already neutralised targets, incorrectly remove surviving threats from the picture or learn the wrong lessons from apparently successful engagements.

Adapt closes the loop. Russia and Ukraine have repeatedly altered frequencies, navigation methods, software, payloads, tactics and countermeasures in response to each other. Adaptation therefore has to include more than tactical improvisation by experienced units. A force must be able to identify why performance changed, communicate the lesson, modify software or procedures, test the change, distribute it and train people before the opponent moves again. NATO’s challenge is particularly demanding because an adaptation may need to propagate across multiple services, commands, suppliers and nations rather than through a single organisation.

Every stage depends on infrastructure that rarely appears in photographs of the platform being celebrated. Communications and networks move information. Spectrum access allows systems to communicate and sense. Positioning, navigation and timing provide location and synchronisation. Data and compute support processing. Power keeps sensors, networks, servers and platforms operating. Software requires configuration, updates and cybersecurity. Interoperability connects systems that were often designed separately. Logistics, maintenance, ammunition, batteries and other consumables keep them available. Trained personnel must understand not only normal operation but failure modes. Doctrine and tactics determine employment. Authorities, rules of engagement and law determine what can be done. Industry must replace losses and replenish consumption. Political decision-making establishes what military commanders are ultimately permitted and required to achieve.

This is the difference between platform-centric and capability-centric planning. A procurement programme can meet its schedule, buy the contracted number of drones and deliver every one to operational units while still failing to produce the intended military effect. A brigade can possess hundreds of drones yet lack sufficient spectrum management, maintainers, batteries, trained crews, data access, integration with fires, replacement capacity or authority to exploit what those drones detect. Conversely, a less impressive platform embedded in a functioning operational chain may produce considerably greater value.

The distinction becomes particularly important when discussing NATO’s proposed Theory of Victory. Counting drones, AI models, sensors or precision weapons tells us what components are available. It does not tell us whether NATO possesses the capability those components are supposed to create. That can only be established by following the chain from sensing through effect and assessment to adaptation, and then asking what happens when the adversary starts breaking its dependencies.

4. 04:00 on Day One

Imagine 04:00 on the eastern flank. The location does not matter for this exercise, and neither does inventing a classified NATO scenario. What matters is the operating environment. An attack is under way or judged imminent. Commanders need to establish what is happening, distinguish genuine activity from deception, allocate forces, protect critical assets and generate effects quickly enough to prevent Russia from establishing momentum. The systems envisaged in NATO’s Theory of Victory now have to perform the functions for which they were designed.

On an architecture diagram, this can look deceptively orderly. Sensors detect activity and feed data into networks. Information is fused into an operational picture. AI-supported tools help classify objects, identify patterns and prioritise information. Commanders make decisions. Tasks move to the appropriate effectors. Drones, aircraft, artillery, missiles, electronic warfare, cyber capabilities and ground forces act. Sensors then assess the results and the process begins again.

At 04:00, however, each arrow on that diagram represents a dependency. The sensor has to be operating and looking in the right place. Its output has to reach a network with sufficient capacity. Positioning, navigation and timing have to remain sufficiently trustworthy for the systems that depend on them. Data formats need to be understood by receiving systems. Security credentials and access permissions have to work. Information from one nation may need to enter another nation’s system. Classification and releasability rules have to permit that exchange. Software versions and interfaces must remain compatible. Computing resources must be available where processing is required, including at the edge when connectivity to remote infrastructure is degraded. Operators need to understand what the systems are telling them and how much confidence to place in the result.

The chain becomes more demanding as it approaches action. A target identified by one sensor may require corroboration from another. The information has to reach the headquarters or unit with authority to act while remaining current enough to be useful. Rules of engagement must cover the situation encountered rather than only the one anticipated. An appropriate effector must be available, within range, supplied and technically able to receive the task. The selected munition, drone or electronic effect must still work in the prevailing environment. After engagement, something has to establish what actually happened. Meanwhile, batteries are being consumed, ammunition stocks are declining, equipment is being damaged and operators are accumulating fatigue.

None of those requirements is particularly exotic. Together, however, they explain why building a digitally enabled multinational fighting force is considerably harder than connecting icons on an architecture slide.

Russia is also unlikely to wait politely while NATO completes the process. Russian doctrine and operations have long incorporated electronic warfare, cyber activity, deception, long-range fires and attacks against command-and-control infrastructure. Ukraine has provided extensive evidence of the interaction. RUSI documented Russian forces using Orlan-10 UAVs, electronic-warfare systems and other sensors to locate Ukrainian positions and feed reconnaissance-strike complexes, while Russian electronic warfare has targeted communications, navigation and drone control. Russia has also repeatedly adapted its own systems when Ukrainian countermeasures reduced their effectiveness.

GNSS illustrates the problem particularly well because modern military and civilian systems have accumulated substantial dependence on reliable positioning and timing. NATO’s 2023 resilience guidance identifies resilient civil communications and transport as requirements for collective defence, while Allied exercises increasingly address operating when space-enabled services are disrupted. Russian interference has already affected GNSS far beyond the Ukrainian battlefield. The European Union Aviation Safety Agency and International Air Transport Association reported a significant increase in GNSS jamming and spoofing affecting aviation and convened joint work specifically because interference had become a broader safety problem.

The same logic applies to communications. More connectivity can increase operational tempo and improve coordination, but connectivity also creates dependencies and signatures. Radio-frequency emissions can reveal locations. Satellite communications can be disrupted. Terrestrial infrastructure can be physically attacked. Fibre offers enormous capacity and resistance to radio-frequency jamming, but a cable still has a physical route that can be cut. Commercial networks may provide valuable redundancy while introducing dependencies on civilian power, infrastructure, operators and supply chains. A resilient architecture therefore cannot merely provide several ways to communicate; it has to understand whether those alternatives fail independently or share the same hidden dependencies.

Cybersecurity adds another layer. The question is not only whether an adversary can disable a system. Compromise can also affect trust in what the system reports. NATO’s 2026 Alliance Digital Strategy explicitly recognises this problem by calling for cybersecurity testing throughout the technology lifecycle, secure hardware and software supply chains, zero-trust approaches, common standards and resilient infrastructure. It also calls for edge computing and communications able to function in degraded, contested and denied environments. These are requirements because an operational network has to remain useful when some of its components cannot be trusted or reached.

The strategy is equally revealing about the less visible integration problems. NATO identifies the need to rationalise legacy networks, implement machine-readable and testable standards, standardise data labelling and metadata, establish federated identity and access management, improve interoperability and reduce dependence on incompatible national systems. Its May 2026 Digital Transformation Implementation Strategy describes the intended Digital Backbone as a federated environment connecting sensors, decision-makers, actors and effectors across national, organisational, operational and security boundaries. It also acknowledges that implementation requires coordination among NATO bodies, national systems and multiple existing initiatives rather than the construction of one new network that simply replaces everything underneath it.

That distinction is important because NATO has not overlooked the problem. The Alliance is actively testing it. Coalition Warrior Interoperability eXploration, eXperimentation, eXamination Exercise, or CWIX, involved more than 570 capabilities and over 25,000 interoperability tests in 2025. In September 2026, Steadfast Cobalt 26 Bravo brought together more than 1,200 personnel from 19 Allies and 37 NATO entities across 41 locations to test and validate national and NATO command-and-control systems for connection to NATO mission networks. These are substantial efforts, but their scale also demonstrates how much work sits behind the apparently simple requirement that Allied systems exchange usable information securely.

The relevant question, therefore, is not whether NATO understands that resilient networks, interoperable data, cybersecurity and digital integration are necessary. Its own strategies, programmes and exercises show that it does. The question is how much of that architecture exists today at operational scale, how consistently it extends into national forces, and how it behaves after several dependencies fail simultaneously rather than sequentially in controlled testing.

That is the difference between connectivity by design and connectivity under attack. At 04:00 on Day One, the architecture does not need every component to work perfectly. It needs to degrade without losing the ability to understand, decide and act. Commanders need to know which functions can continue locally, which require reach-back, which alternative communications paths are genuinely independent, what happens when trusted data becomes uncertain, and how units operate when the digital services on which they trained are temporarily unavailable.

Russia does not have to destroy every NATO sensor, drone or command post to interfere with the theory of victory. It only has to find dependencies whose failure interrupts enough of the chain for long enough to create operational advantage. The measure of the architecture is therefore not how well everything connects when the system is intact. It is how much useful military capability remains when parts of it are jammed, spoofed, compromised, disconnected, destroyed or simply unavailable at the moment they are needed.

5. The distance between an architecture and an Alliance

NATO’s digital challenge has a characteristic that no individual military service or technology company can remove: the Alliance consists of 32 sovereign nations. They do not arrive at an operation as components of one centrally designed military system. They bring different generations of equipment, communications networks, security classifications, data policies, acquisition systems, doctrines, training standards, cyber requirements, industrial suppliers, legal frameworks and political authorities. They also bring different levels of digital maturity and different national restrictions on how capabilities and information may be used. NATO has to make military effect from that diversity without eliminating the sovereignty that created it.

Federation is therefore both a strength and an integration problem. It allows Allies to retain national forces, industries, infrastructure and decision-making while combining them for collective defence. It also means that an operational chain may cross several institutional boundaries before producing an effect. A sensor owned by one nation may generate information processed through a NATO service, enriched with intelligence from another Ally and eventually used to task an effector operated by a fourth. Each transition can introduce technical, procedural, security, legal or political conditions that do not appear when the individual systems are examined separately.

The first layer is technical interoperability: whether systems can physically and digitally exchange information. Interfaces, protocols, data formats, network connectivity, software versions, encryption and identity management all matter. This is already difficult when military equipment remains in service for decades. A newly procured system may need to communicate with a platform designed before its current operators entered military service. NATO’s 2026 Alliance Digital Strategy consequently calls for rationalising legacy networks, greater use of open standards and modular architectures, common services, federated identity and access management, and machine-readable and testable standards. Those are not administrative preferences. They determine whether information can travel through the operational system without expensive bespoke integration at every boundary.

Successful transmission does not establish semantic interoperability. Two systems can exchange a message while interpreting its contents differently. Track identifiers, confidence levels, target categories, geographic references, timestamps and operational terminology have to carry sufficiently consistent meaning. Even seemingly straightforward terminology becomes consequential when humans and machines use it to make decisions. NATO standardisation has addressed common terminology and procedures for decades, but digital operations increase the requirement because machines cannot resolve ambiguity through the informal conversations humans routinely use to clarify meaning. NATO’s Data Strategy therefore places emphasis on metadata, standards, data quality and shared data spaces rather than treating data exchange as a matter of network capacity alone.

Procedural interoperability asks whether forces know how to use the exchanged information together. A common operational picture has limited value if participating units use different processes for validation, targeting, reporting or battle-damage assessment. Training becomes important here because interoperability cannot be produced entirely through documentation. NATO exercises such as CWIX test technologies and standards, while operational exercises expose the interaction between systems, procedures and people. The more NATO seeks to compress decision and engagement timelines, the less room exists for manual reconciliation between national processes after information has already arrived.

Operational interoperability goes further by asking whether the combined system can produce the intended military effect. Two national systems may exchange information correctly and their personnel may follow compatible procedures, yet the information can still arrive too slowly for the target, lack the precision required by the selected effector or omit information necessary for the commander to authorise engagement. Interoperability should therefore be measured against mission outcomes rather than successful connections. A green status light on an interface test is useful; it does not demonstrate that a multinational sensor-to-effector chain can find and engage a fleeting target under battlefield conditions.

Organisational interoperability concerns responsibilities, command relationships and institutional boundaries. NATO forces operate through established command structures, but digital information can move across organisational boundaries much faster than traditional processes were designed to accommodate. A sensor may provide information almost instantaneously while the authority required to exploit it remains elsewhere. Adding AI-supported prioritisation does not remove that constraint. If anything, faster technology exposes organisational delay more clearly because the machine can produce a recommendation long before the institution is ready to act on it.

Then come legal and political interoperability, which technology cannot engineer away. Data may be technically transferable while national law, classification rules, intelligence-handling restrictions or bilateral agreements prevent its release. A weapon may be technically capable of an effect that national rules of engagement do not permit. Cyber and electronic-warfare operations can create effects beyond a commander’s immediate area and interact with civilian infrastructure or another state’s networks, increasing legal and political sensitivity. National caveats have historically affected NATO operations because Allies retain authority over their forces. In a highly connected force, such restrictions can also influence how information and machine-generated recommendations move through the system.

These layers explain why NATO’s 2025 Data Strategy is significant. It sets a 2030 objective for an Alliance Data Sharing Ecosystem connected through the NATO Digital Backbone, supported by curated quality data, optimised data-management processes and a data-literate workforce. The strategy explicitly envisages movement away from isolated repositories towards shared data spaces and federated data meshes, while recognising that data ownership, stewardship, standards, security and workforce skills have to develop alongside the technical architecture. The 2030 date should not be treated as evidence of failure today. It is evidence that NATO itself regards this as a multi-year transformation rather than a network upgrade that can simply be switched on.

There is also an important distinction between achieving interoperability among selected systems during an exercise and maintaining it across a force that is continuously changing. Nations buy new equipment on different schedules. Software is updated. Security vulnerabilities require patches. Classification policies change. New suppliers enter architectures. Lessons from Ukraine alter requirements. Mobilisation can introduce reserve formations and older equipment that were not central to peacetime integration efforts. Interoperability is therefore not a certification obtained once. It is a condition that has to be maintained while every component around it continues to evolve.

For the Theory of Victory, this creates a leadership problem as much as a technical one. Commanders and policymakers need to know where information and authority can move at operational speed, where national boundaries introduce delay, which systems remain dependent on manual workarounds and which restrictions require political decisions before a crisis begins. Procurement leaders need to treat integration requirements as part of capability acquisition rather than work deferred until platforms arrive.

Thirty-two nations do not need identical equipment or identical institutions to fight together. They do need sufficient technical, semantic, procedural, operational, organisational, legal and political interoperability for the required military effect to cross the boundaries between them. If a sensor detects the right target and the network successfully transmits the data, but the receiving force cannot interpret, release, trust or legally act upon it, the technology has performed exactly as designed while the Alliance capability has still failed.

6. Tested somewhere is not ready for war

Defence technology is routinely described with terms that sound more precise than they are. A system has been demonstrated, tested, validated, fielded or described as combat-proven, and those labels can gradually become substitutes for asking what evidence actually exists. For a Theory of Victory that depends on large numbers of interconnected systems functioning under deliberate attack, the distinction is operationally important. A successful demonstration can justify further development. It cannot by itself justify an assumption that the capability will work at Alliance scale in war.

For this series, demonstrated means that a technology or function has been shown to work under stated conditions. A drone may complete an autonomous flight, an algorithm may classify imagery, or two command systems may exchange information. The demonstration establishes possibility. Its value depends on knowing the conditions: what was controlled, what support was available, what threats were absent and what failures occurred alongside the successful result.

Tested means performance has been deliberately examined against defined criteria. Testing should expose limits rather than merely reproduce success. Range, weather, latency, accuracy, electromagnetic interference, cyber resilience, operator workload and failure behaviour can all be measured, but the test design determines what the result proves. Testing an uncrewed system in benign spectrum conditions does not establish its performance under sophisticated jamming. Testing an AI classifier against data drawn from the same environment that produced its training data does not establish how it will behave when battlefield conditions change.

Validated requires stronger evidence that the system performs its intended function under conditions sufficiently representative of its intended operational use. This is where mission context becomes unavoidable. NATO’s revised AI strategy explicitly calls for an Alliance-wide testing, evaluation, verification and validation, or TEV&V, landscape and says AI should be assessed under different technological, procedural and operational conditions. NATO also identifies adversarial AI, data quality, bias, explainability and changes in system behaviour as issues requiring lifecycle attention rather than one-time approval.

Operationally fielded means that forces actually possess and employ the system. That is a substantial step beyond experimentation, but fielding still says little about scale or consistency. A specialist unit with experienced operators, direct developer support and priority access to communications or intelligence may achieve results that ordinary formations cannot reproduce. Fielding therefore has to be examined alongside who operated the system, how they were trained, what support they received and what happened when equipment failed.

Integrated means the system works inside the wider operational chain rather than as an isolated product. A sensor has to provide usable information to command systems. Command systems have to support decisions and tasking. Effectors need to receive and execute those tasks. Identity, security, data standards and authorities have to survive every transition. NATO’s CWIX exists precisely because this cannot be assumed from individual system performance. The 2025 event involved more than 570 capabilities and over 25,000 interoperability tests across areas including command and control, space, cyber, joint fires and digital interoperability.

Scalable asks whether an effect demonstrated by a few systems and highly experienced people can be reproduced at the volume war requires. Scaling changes the problem. Thousands of drones require thousands of batteries, payloads, replacement parts and communications channels, together with operators, maintainers, instructors, storage, transport and spectrum management. AI services require compute, data pipelines, cybersecurity and people capable of monitoring their performance. What works for ten systems may encounter completely different constraints at ten thousand.

Sustainable adds time. Can the capability continue after equipment is lost, components become scarce, software needs updating, operators become casualties or exhausted, infrastructure is attacked and the enemy develops countermeasures? Industrial production, repair capacity, ammunition, training pipelines, software support and access to critical components become part of the answer. A system can therefore be demonstrated, tested, validated, fielded, integrated and scalable while still proving unsustainable in a prolonged war.

The environment in which each claim is established matters just as much. A laboratory allows variables to be isolated and controlled. A range introduces physical operating conditions. An exercise can test interactions among people, organisations and systems. Ukraine provides evidence from actual combat against an adaptive opponent. Each environment answers questions the others cannot, but none is a universal proxy for the next war. Ukrainian experience reflects particular terrain, force structures, authorities, infrastructure, industrial arrangements, spectrum conditions and an adversary that has itself changed substantially since 2022.

NCIA’s Counter-UAS Data Challenge, launched in September 2026, provides a useful current example of how difficult even the beginning of the chain remains. NATO is seeking approaches that combine radar, radio-frequency, camera and acoustic information to improve the detection, tracking, classification and identification of small drones. The problem is difficult because individual sensors provide incomplete or ambiguous information and because small drones operate in environments containing birds, civilian activity, clutter, friendly systems and deliberate attempts at deception.

That is considerably removed from the simplified proposition that AI identifies a hostile drone and an autonomous system destroys it. Before an engagement decision exists, somebody has to establish whether the object was detected correctly, whether several sensors are observing the same object, what it probably is, how confident that assessment is and what happens when the sensors disagree. Those are not peripheral technical details. They determine whether the rest of the operational chain should begin at all.

The evidence standard for this series is therefore deliberately demanding. A technology that worked somewhere in Ukraine is evidence worth examining, not a blank cheque marked “combat-proven.” The relevant questions are where it worked, for how long, against what opposition, with which software and support, how often it failed, whether another unit reproduced the result, what happened when the adversary adapted, and whether NATO can integrate and sustain the same effect at the scale required for collective defence.

7. What if Day One becomes Year Four?

A theory of victory understandably concentrates on how NATO intends to prevail. It also has to account for what happens when the adversary refuses to cooperate with the expected timetable. Russia’s full-scale war against Ukraine is now well into its fifth year. Technologies that appeared decisive at different stages of the conflict have been countered, modified and reintroduced in new forms, while requirements for ammunition, personnel, air defence, electronic warfare, drones, repair and industrial production have accumulated rather than disappeared. Technological superiority can create operational advantage. It does not determine how long an opponent is willing or able to continue fighting.

This is where the distinction between a theory of victory and a theory of war becomes important. NATO may successfully stop an initial Russian advance and destroy substantial elements of the attacking force without producing immediate war termination. Russia could continue long-range strikes, regenerate forces, mobilise additional personnel, attack infrastructure, conduct cyber and information operations, or seek opportunities elsewhere along the Alliance’s perimeter. NATO could therefore win the opening campaign and still face the harder strategic problem of determining what conditions would actually compel Moscow to stop.

Iraq and Afghanistan are useful here only if their differences from an Article 5 war against Russia are kept explicit. Neither provides a template for collective defence in Europe. Both do, however, demonstrate the consequences of allowing preferred assumptions about duration, objectives and political outcomes to shape planning.

The UK’s Iraq Inquiry found that preparations for the aftermath of the 2003 invasion were inadequate despite identifiable risks. In presenting the inquiry’s findings to Parliament, Prime Minister David Cameron noted that British planning had remained too dependent on assumptions that the United States had an adequate plan, that the United Nations and wider international community would share the burden, and that the UK role would end approximately three to four months after the initial conflict. The inquiry concluded that inadequate preparation for the aftermath reduced the likelihood of achieving Britain’s strategic objectives.

Afghanistan presents a different warning. NATO’s post-2021 lessons process concluded that the international community’s ambition had expanded far beyond the original objective of degrading terrorist safe havens. Among its recommendations were continuous reassessment of strategic interests, achievable goals and greater awareness of mission expansion. NATO had demonstrated enormous military capacity during two decades of operations, but military capacity did not by itself resolve the political problem the intervention had gradually acquired.

The transferable lesson is not that a future war with Russia would resemble Iraq or Afghanistan. It is that planning around the preferred duration and preferred outcome leaves known risks outside the plan. For collective defence, those risks become particularly serious because NATO would be sustaining operations across 32 nations while protecting civilian populations and infrastructure, maintaining deterrence elsewhere, supporting economic activity and managing escalation with a nuclear-armed adversary.

The first question is therefore what happens if Russia is stopped but does not terminate the war. A 30-day campaign and a four-year conflict require different military systems. Personnel need rotation and replacement. Damaged units require regeneration. Experienced operators become casualties or instructors. Equipment designed around peacetime maintenance schedules begins accumulating operating hours at wartime rates. Training establishments have to replace skills while simultaneously incorporating battlefield lessons. Infrastructure needs repair while remaining under threat.

Industrial assumptions change just as dramatically. “Cheap mass” remains cheap only at the level of the individual unit. Producing millions of drones means securing millions of motors, processors, batteries, cameras, antennas and other components, alongside explosives, production facilities, skilled labour, transport and increasingly large software and data-support systems. Sophisticated air-defence interceptors, long-range missiles and precision munitions create different but equally consequential production constraints. NATO’s updated Defence Production Action Plan now explicitly calls for understanding industrial capacity in peacetime, crisis and conflict, identifying surge options, protecting defence-critical supply chains and assessing whether production can be sustainable and scalable over the long term.

The Alliance is therefore already addressing a problem that Ukraine has made impossible to ignore. Increased defence budgets do not automatically create wartime output. NATO acknowledged in July 2026 that industrial capacity has to expand alongside expenditure, and Secretary General Mark Rutte stated when launching the NATO Engine that no single Ally possesses the industrial capacity required to meet growing demand, particularly for strike and air-defence capabilities.

Time also changes technology itself. What if an AI-enabled targeting advantage lasts six weeks before Russia identifies its weakness? What if a drone that performs exceptionally today becomes increasingly ineffective after frequencies are mapped, signatures identified or countermeasures distributed? Procurement processes designed to select a product and operate it for years sit uneasily beside capabilities whose software, communications methods and tactics may need modification within weeks. Sustaining technological advantage consequently means sustaining an adaptation system, not merely sustaining inventories.

The political problem develops on the same timeline. Governments change. Elections occur. Defence expenditure competes with other public priorities. Infrastructure damage affects civilian life. Casualties accumulate. Industrial mobilisation imposes costs. Disinformation and influence operations seek to exploit disagreements among Allies. A four-year war therefore tests political cohesion and public endurance alongside military readiness. Those are not peripheral considerations because NATO operates by political agreement among sovereign democracies.

A credible Theory of Victory must consequently survive several uncomfortable questions. What if losses exceed the assumptions used to size the force? What if Russia regenerates faster than expected? What if ammunition rather than platforms becomes the limiting factor? What if technological advantages repeatedly expire? What if NATO achieves military superiority locally but cannot produce acceptable conditions for war termination? What level of industrial, personnel and political mobilisation can the Alliance sustain if the answer to “how long?” is measured in years?

These questions belong inside deterrence rather than outside it. An adversary considering aggression is not only assessing whether NATO can defeat the opening attack. It is also assessing whether the Alliance can absorb losses, replace what it consumes, adapt when its first solutions are countered and maintain political purpose after the original plan encounters friction. A Theory of Victory becomes more credible when it explains not only how NATO intends to win quickly, but why failing to win quickly would not prevent NATO from continuing until it does.

8. The questions a Theory of Victory must survive

A Theory of Victory cannot answer every operational question in an unclassified document, nor should it reveal how NATO intends to fight in sufficient detail to help an adversary. It should, however, be possible to test the assumptions underneath it. The purpose of the following questions is therefore not to demand public disclosure. It is to identify where an unanswered question becomes operational risk and where work has to be completed before a crisis rather than during one.

Military outcome. What constitutes victory, what conditions terminate the conflict, and what happens if Russia refuses those conditions? These questions establish what military force is ultimately supposed to accomplish. Halting an incursion, restoring Allied territory, degrading Russian forces and compelling Moscow to stop fighting require overlapping but different capabilities and levels of commitment. Without a sufficiently clear political end state, operational success can accumulate without resolving the strategic problem. The relevant test is whether military objectives create conditions for an achievable political outcome rather than simply producing a favourable battlefield position.

Operational architecture. Which functions must remain connected, which can degrade gracefully, and which must continue when completely disconnected? A force designed around permanent access to data, PNT, cloud services or high-bandwidth communications carries vulnerabilities if those dependencies are not explicit. Commanders need to know what happens when information arrives late, when headquarters cannot be reached, when sensors disagree or when units have to operate locally with incomplete information. Resilience is therefore partly the ability to continue producing useful effects at reduced performance rather than preserving perfect connectivity.

Technology. What has been demonstrated, tested, validated, operationally fielded, integrated and scaled, and under what conditions? The distinction determines how much operational confidence can reasonably be placed in a system. Leaders also need to know its failure modes, what support produced successful results and what happened after an adversary adapted. A technology portfolio containing many successful demonstrations can still contain very little deployable capability.

People. Who operates, understands, repairs, challenges and adapts these systems, and how quickly can those people be replaced? Digital and autonomous systems can reduce some personnel requirements while creating others in software, data, cyber security, maintenance, spectrum management and technical command functions. Training operators to use a system is only one requirement. Forces also need people capable of recognising when its output is unreliable, diagnosing failures and changing procedures when battlefield conditions invalidate previous assumptions.

Authority. Who can decide, share, task and engage, and within what timeframe? A sensor-to-effector chain that technically operates in seconds gains little if information requires lengthy release procedures or engagement authority sits several organisational levels away. Delegation, rules of engagement, national caveats and legal authorities therefore need to match the tempo the architecture is designed to achieve. Where they cannot, the delay should be understood as part of capability performance rather than treated as an external policy problem.

Data. Who owns the data, who labels and validates it, who can access it, who protects it, and who determines whether it remains trustworthy? NATO’s Data Strategy treats quality, stewardship, standards, security and workforce literacy as prerequisites for data exploitation precisely because quantity alone is insufficient. AI makes this more consequential: poor, outdated, manipulated or incorrectly labelled data can produce machine outputs that appear precise while resting on weak evidence.

Industry. How quickly can destroyed and consumed systems be replaced, and which dependencies determine the real production rate? The answer is not simply the number of final platforms a factory can assemble. Batteries, processors, motors, antennas, optics, explosives, cables, specialist materials, software, skilled labour and secure supply chains can each become the constraint. Production capacity has operational meaning only when the entire supply chain can sustain the required output.

Alliance. Which dependencies require agreement among 32 sovereign states before the crisis begins? Data-sharing arrangements, network access, national caveats, host-nation support, cross-border movement, authorities and common standards are difficult to negotiate while forces are already trying to fight. NATO does not need every national rule to be identical, but it needs to know beforehand where differences interrupt the operational chain and what mechanisms resolve them.

Adversary. Which assumptions can Russia attack most cheaply? The most dangerous vulnerability may not be the most sophisticated system. It may be a communications node, commercial component, power supply, software dependency, data source, maintenance facility or political disagreement whose disruption creates effects elsewhere in the architecture. Red-teaming should therefore look for disproportionate consequences rather than merely technically impressive attacks.

Time. Finally, what changes when a 30-day concept becomes a 300-day campaign or a four-year war? Stocks decline, equipment wears, personnel rotate, adversaries adapt, software changes, governments change and industrial capacity becomes increasingly important. A Theory of Victory has to remain coherent when the timeline extends beyond the scenario its planners would prefer.

These questions do not require perfect answers. They require identified assumptions, responsible owners, evidence and contingencies. What remains unanswered in peacetime does not disappear in war. It usually returns under worse conditions, with less time available to solve it.

9. What Part 1 leaves us with

NATO’s emerging Theory of Victory deserves to be taken seriously because the military problem it addresses has changed materially. Russia can generate mass, absorb losses and adapt. Ukraine has demonstrated the operational value of inexpensive uncrewed systems, distributed sensing, rapid software and tactical adaptation, and shorter links between detection and effect. NATO therefore needs affordable mass alongside exquisite capabilities, greater use of autonomy and AI-assisted processing, resilient distributed sensing and forces able to modify how they fight faster than traditional acquisition and capability-development cycles have generally allowed.

None of that removes the requirement for conventional military power. Artillery, air and missile defence, long-range fires, aircraft, armour, infantry, engineering, logistics, electronic warfare, cyber capabilities and nuclear deterrence remain part of the force that NATO expects to deter and, if necessary, defeat aggression. The more useful question is how emerging technologies change the way those capabilities find, understand and engage an adversary, and whether they can be connected into a system that continues to function when parts of it fail.

NATO’s own strategies show that the Alliance understands much of the work underneath that ambition. The Alliance Digital Strategy addresses legacy networks, interoperability, edge computing, cyber security, common standards and resilient infrastructure. The Data Strategy identifies data quality, stewardship, sharing and workforce literacy as prerequisites for exploiting data and AI. The Defence Production Action Plan addresses production capacity, long-term demand, supply-chain security and the ability to scale in crisis and conflict. These are different policy areas, but operationally they meet in the same place: a military capability exists only when technology, people, infrastructure, authorities, data, logistics and industry can produce the required effect together.

The resources are also changing. At the 2025 Hague Summit, Allies committed to invest 5% of GDP annually in defence and security-related requirements by 2035, with at least 3.5% directed towards core defence requirements and NATO capability targets and up to 1.5% towards areas including critical infrastructure, networks, resilience, innovation and the defence industrial base. More money creates options, but expenditure is not itself capability. What is purchased, how it is integrated, whether people can operate it and whether industry can sustain it will determine how much additional combat power those commitments ultimately produce.

The Theory of Victory therefore gives NATO a useful direction of travel, while implementation determines whether that direction becomes executable warfighting capability. The next problem follows directly from the architecture described in this article. Much of its effectiveness depends on information moving between sensors, decision-makers and effectors quickly enough to matter. Russia does not have to destroy every NATO platform individually if it can disrupt the communications, spectrum, positioning and timing, data flows, infrastructure or trust relationships that allow those platforms to operate as a coherent force. Part 2 therefore moves from the architecture NATO wants to build to the environment in which it would actually have to operate: a battlefield actively trying to disconnect it.

Upcoming in the series

Part 2- The Battlefield Is Trying to Disconnect You

The connected battlespace works only for as long as its dependencies survive. Part 2 examines electronic warfare, GNSS jamming and spoofing, communications, tactical data links, SATCOM, fibre, 5G and future networks, power, cyber attack and infrastructure. It asks how NATO should design for degradation rather than assume persistent connectivity, and what military capability remains when parts of the network are unavailable, compromised or deliberately denied.

Part 3- AI Does Not Know What It Is Looking At

AI can process more information faster than humans, but speed does not establish truth. Part 3 examines target recognition, sensor fusion, confidence levels, false positives, training data, data labelling, poisoning, adversarial manipulation, hallucination, automation bias and the limits of human oversight at machine speed. It asks how commanders know when a system is wrong, what evidence should be required before AI influences an engagement decision, and who is accountable when machine-supported judgement fails.

Part 4-The Integration War Nobody Photographs

The least glamorous work may determine whether the most advanced systems function at all. Part 4 examines APIs, gateways, legacy platforms, software baselines, identity management, classifications, data models, taxonomies, cross-domain solutions, accreditation, interoperability standards and national restrictions. It asks whether NATO is still buying platforms and attempting to integrate them afterwards when capability should instead be designed around the operational chain from the beginning.

Part 5- Who Is Allowed to Press the Button?

Technology can accelerate a decision without resolving who has the authority to make it. Part 5 examines command responsibility, rules of engagement, national caveats, legal constraints, meaningful human control, accountability, AI governance, testing, bias, explainability, political authority and the cultural consequences of introducing increasingly autonomous systems into military decision-making. It also asks what human control actually means when the volume and speed of machine-generated recommendations exceed the capacity for meaningful human review.

Part 6- From Theory of Victory to a System That Can Fight

The final part turns the preceding problems into requirements. It examines what NATO would need across mission architecture, resilient connectivity, data, interoperability, TEV&V, authorities, personnel, industry, sustainment, cyber security, exercises, procurement, adaptation and political decision-making to turn the Theory of Victory into executable warfighting capability. It returns to the question running through the series: what must remain true not only for NATO to win quickly, but for the Alliance to continue fighting, adapting and regenerating if it does not?

Materials and sources used for Part 1

NATO’s Theory of Victory and the emerging concept

Bloomberg- NATO Drafts Plan to Defeat a Russian Attack With Drones and AI 5 October 2026 Original reporting that prompted this series and reported the unclassified Theory of Victory being developed under SACEUR General Alexus Grynkewich. https://www.bloomberg.com/news/articles/2026-10-05/nato-drafts-plan-to-defeat-a-russian-attack-with-drones-and-ai

ANSA- Here is NATO’s Theory of Victory, designed to defeat Moscow on its eastern flank 5 October 2026 Additional reporting on the sequence described in the unclassified document, including halting an incursion, countering mass and momentum through the Eastern Flank Deterrence Initiative, attacking C5ISRT and subsequently defeating the wider combat system. https://www.ansa.it/europa/en/news/2026/10/05/here-is-natos-theory-of-victory-designed-to-defeat-moscow-on-its-eastern-flank_4b6bcc65-48a0-48b8-838c-39bcdbfbf329.html

NATO deterrence, defence and digital transformation

NATO- Deterrence and Defence Background on NATO’s overall deterrence and defence posture, including the combination of conventional, nuclear and missile-defence capabilities complemented by space and cyber capabilities. https://www.nato.int/en/what-we-do/deterrence-and-defence/deterrence-and-defence

NATO- Alliance Digital Strategy 13 January 2026 Central source for NATO’s digital transformation, including the Digital Backbone, resilient and interoperable infrastructure, legacy-network rationalisation, Software Defined Defence, edge computing, Zero Trust, federated platforms, data-centric operations, AI, standardised data labelling, identity and access management, workforce requirements and operations in degraded, contested and denied environments. https://www.nato.int/en/about-us/official-texts-and-resources/official-texts/2026/01/13/alliance-digital-strategy

NATO- Digital Transformation Implementation Strategy 26 May 2026 Implementation framework for NATO’s digital transformation and the federated Digital Backbone connecting sensors, decision-makers, actors and effectors across national, organisational, operational and security boundaries. https://www.nato.int/en/about-us/official-texts-and-resources/official-texts/2026/05/26/natos-digital-transformation-implementation-strategy

NATO- Data Strategy for the Alliance 5 May 2025 NATO’s 2030 data target and the development of the Alliance Data Sharing Ecosystem, including data quality, curation, governance, data spaces, federated approaches, interoperability and workforce literacy. https://www.nato.int/en/about-us/official-texts-and-resources/official-texts/2025/05/05/data-strategy-for-the-alliance

NATO- Summary of NATO’s Revised Artificial Intelligence Strategy 10 July 2024 NATO’s approach to responsible AI, interoperability, AI-enabled capabilities, adversarial AI and the requirement for testing, evaluation, verification and validation across different technological, procedural and operational conditions. https://www.nato.int/en/about-us/official-texts-and-resources/official-texts/2024/07/10/summary-of-natos-revised-artificial-intelligence-ai-strategy

Interoperability, testing and Counter-UAS

NATO Allied Command Transformation- CWIX 25 Concludes Reporting on Coalition Warrior Interoperability eXploration, eXperimentation, eXamination Exercise 2025, involving more than 570 capabilities and over 25,000 interoperability tests. https://www.act.nato.int/article/cwix25-concludes/

NATO Communications and Information Agency- NATO launches data challenge to counter drone threats September 2026 NCIA’s Counter-UAS Data Challenge examining the fusion of radar, radio-frequency, camera and acoustic data for detecting, tracking, classifying and identifying small drones. https://www.ncia.nato.int/newsroom/news/nato-launches-data-challenge-to-counter-drone-threats

NATO Communications and Information Agency- NCIA supports mission network expansion during Steadfast Cobalt 26 Bravo September 2026 Current example of NATO testing and validating national and NATO command-and-control systems for connection to mission networks across 19 Allies, 37 NATO entities and 41 locations. https://www.ncia.nato.int/newsroom/news/-ncia-supports-mission-network-expansion-during-steadfast-cobalt-26-bravo-

Ukraine, drones, adaptation and combined arms

Ukraine Ministry of Defence- By the end of the year, the Armed Forces will receive 3 million FPV drones 24 December 2025 Official Ukrainian figures illustrating the industrial scale that FPV warfare has reached. https://mod.gov.ua/news/do-kinczya-roku-zsu-otrimayut-3-miljoni-odinicz-fpv-droniv-denis-shmigal

Ukraine Ministry of Defence- Systemic changes to strengthen Ukraine’s military: Denys Shmyhal reports to parliament on his tenure as Minister of Defence 2026 More recent figures on Ukrainian drone delivery, including FPVs, fibre-optic drones, deep-strike systems, reconnaissance drones and interceptor drones, together with procurement and delivery reforms. https://mod.gov.ua/en/news/systemic-changes-to-strengthen-ukraine-s-military-denys-shmyhal-reports-to-parliament-on-his-tenure-as-minister-of-defense

CSIS- Why Ukraine Is Establishing Unmanned Forces 2024 Analysis of Ukraine’s institutional response to rapidly evolving unmanned warfare and the speed of battlefield adaptation. https://www.csis.org/analysis/why-ukraine-establishing-unmanned-forces

CSIS- Ukraine’s Future Vision and Current Capabilities for Waging AI-Enabled Autonomous Warfare Analysis of Ukrainian experimentation with AI, autonomous and semi-autonomous functions, navigation, target recognition and the operational pressures driving greater autonomy. https://www.csis.org/analysis/ukraines-future-vision-and-current-capabilities-waging-ai-enabled-autonomous-warfare

CSIS- Technological Evolution on the Battlefield Analysis of drones, ISR, shortened targeting cycles, commercial technology, procurement adaptation and the changing relationship between technology and battlefield signatures. https://www.csis.org/analysis/chapter-9-technological-evolution-battlefield

CSIS- Shahed/Geran: How Russia Continues to Reinvent the One-Way Attack Drone Analysis of Russia’s continuous modification of the Shahed/Geran family, including navigation, communications, payload, propulsion and operational adaptation. https://www.csis.org/analysis/shahed-geran-how-russia-continues-reinvent-one-way-attack-drone

RUSI- Meatgrinder: Russian Tactics in the Second Year of Its Invasion of Ukraine Detailed field research on Russian reconnaissance-strike complexes, UAVs, electronic warfare, artillery, command structures and tactical adaptation. https://www.rusi.org/explore-our-research/publications/special-resources/meatgrinder-russian-tactics-second-year-its-invasion-ukraine

RUSI- Emergent Approaches to Combined Arms Manoeuvre in Ukraine Field-based examination of the interaction between drones, artillery, electronic warfare, infantry and other battlefield capabilities, including environmental and tactical limitations on unmanned systems. https://www.rusi.org/explore-our-research/publications/insights-papers/emergent-approaches-combined-arms-manoeuvre-ukraine

RUSI- NATO Should Not Replace Traditional Firepower with Drones Analysis of why lessons from Ukraine support integration of drones with conventional firepower rather than treating uncrewed systems as wholesale replacements for traditional capabilities. https://www.rusi.org/explore-our-research/publications/rusi-defence-systems/nato-should-not-replace-traditional-firepower-drones

GNSS, infrastructure and contested connectivity

European Union Aviation Safety Agency and IATA- Comprehensive plan to mitigate GNSS interference risks 18 June 2025 Evidence of the growing scale of GNSS jamming and spoofing beyond the Ukrainian battlefield and the requirement for resilience and alternative navigation arrangements. https://www.easa.europa.eu/en/newsroom-and-events/press-releases/easa-and-iata-outline-comprehensive-plan-mitigate-gnss

EASA and EUROCONTROL- Joint Action Plan to ensure safe operations during GNSS interference events 25 March 2026 Updated European work on operational resilience as GNSS interference becomes a regular feature around conflict zones. https://www.easa.europa.eu/en/newsroom-and-events/press-releases/easa-and-eurocontrol-publish-joint-action-plan-ensure-safe

Strategic assumptions, duration and war termination

UK Government | Prime Minister’s statement on the Iraq Inquiry 6 July 2016 Summary of the Chilcot Inquiry’s findings, including shortcomings in planning for the aftermath of the invasion and assumptions concerning the expected duration of the British role. https://www.gov.uk/government/speeches/pm-statement-on-the-iraq-inquiry-6-july-2016

NATO- NATO and Afghanistan NATO’s account of the Afghanistan mission and its lessons-learned process, including conclusions concerning mission expansion, continuous reassessment of strategic interests and the need to establish achievable objectives. https://www.nato.int/en/what-we-do/operations-and-missions/nato-and-afghanistan

Industrial capacity and defence investment

NATO- Updated Defence Production Action Plan 13 February 2025 NATO framework addressing defence-industrial capacity, long-term demand, production in peacetime, crisis and conflict, supply-chain security, surge requirements and sustainable production. https://www.nato.int/en/about-us/official-texts-and-resources/official-texts/2025/02/13/updated-defence-production-action-plan

NATO- Defence investment and NATO’s 5% commitment, updated 29 June 2026 Explanation of the Hague commitment to invest 5% of GDP annually by 2035, including at least 3.5% for core defence requirements and NATO Capability Targets and up to 1.5% for broader defence and security-related requirements. https://www.nato.int/en/what-we-do/introduction-to-nato/defence-expenditures-and-natos-5-commitment

NATO- The Hague Summit Declaration 25 June 2025 Primary political source for the 5% commitment and the linkage between investment, forces, capabilities, infrastructure, readiness and resilience. https://www.nato.int/en/about-us/official-texts-and-resources/official-texts/2025/06/25/the-hague-summit-declaration

Suggested further reading

The following materials provide useful background beyond the sources directly relied upon in Part 1 and will also support later articles in this series.

NATO strategy, AI, data and emerging technology

NATO- 2022 Strategic Concept The Alliance’s overarching strategic framework, including Russia, deterrence and defence, resilience, technological advantage, cyber and space. https://www.nato.int/strategic-concept/

NATO- Summary of the NATO Artificial Intelligence Strategy 22 October 2021 The original framework establishing NATO’s Principles of Responsible Use for AI, including lawfulness, responsibility and accountability, explainability and traceability, reliability, governability and bias mitigation. https://www.nato.int/cps/en/natohq/official_texts_187617.htm

NATO- Emerging and Disruptive Technologies Background on NATO’s wider work across AI, autonomy, quantum technologies, biotechnology, space and other emerging technologies. https://www.nato.int/cps/en/natohq/topics_184303.htm

NATO interoperability and standardisation

NATO- Interoperability: Connecting Forces Background on why technical, procedural and human interoperability is central to multinational NATO operations. https://www.nato.int/cps/en/natohq/topics_84112.htm

NATO Standardization Office Background on NATO standardisation and the role of standards in enabling Allied forces to operate together. https://www.nato.int/cps/en/natohq/topics_124879.htm

Ukraine and battlefield adaptation

RUSI- Preliminary Lessons in Conventional Warfighting from Russia’s Invasion of Ukraine: February–July 2022 One of the useful early studies for understanding how the initial assumptions of the war collided with battlefield realities, including fires, logistics, EW, command and force employment. https://www.rusi.org/explore-our-research/publications/special-resources/preliminary-lessons-conventional-warfighting-russias-invasion-ukraine-february-july-2022

RUSI- Stormbreak: Fighting Through Russian Defences in Ukraine’s 2023 Offensive

Detailed analysis of the difficulties of combined-arms manoeuvre against prepared defences, mines, artillery, UAVs, electronic warfare and layered Russian positions. https://www.rusi.org/explore-our-research/publications/special-resources/stormbreak-fighting-through-russian-defences-ukraines-2023-offensive

RUSI- Tactical Developments During the Third Year of the Russo-Ukrainian War Further battlefield analysis of the continuing adaptation of Russian and Ukrainian tactics and the interaction among drones, fires, EW, protection and manoeuvre. https://www.rusi.org/explore-our-research/publications/special-resources/tactical-developments-during-third-year-russo-ukrainian-war

War, strategy and the danger of assumptions

The Iraq Inquiry- Executive Summary Primary findings of the UK Iraq Inquiry concerning decision-making, military planning, preparation for the aftermath and the relationship between strategic objectives and available resources. https://www.gov.uk/government/publications/the-report-of-the-iraq-inquiry

NATO- Afghanistan Lessons Learned The lessons section within NATO’s Afghanistan overview remains particularly relevant for thinking about strategic objectives, mission expansion, political assumptions and the difference between military activity and sustainable strategic outcomes. https://www.nato.int/en/what-we-do/operations-and-missions/nato-and-afghanistan

Series principle

Throughout this series, terms such as demonstrated, tested, validated, operationally fielded, integrated, scalable and sustainable are used deliberately. Evidence that a technology worked in one exercise, laboratory, range or sector of the Ukrainian battlefield is valuable, but it does not establish that the same capability can be reproduced across a multinational NATO force, under deliberate electronic and cyber attack, at the scale and for the duration required by collective defence.

The same standard applies to the Theory of Victory itself. The purpose of this series is not to argue against the direction NATO is taking. It is to examine what must be built, changed, tested, integrated, authorised, supplied and sustained for that direction to become executable warfighting capability.