The shape of modern warfare: How defence technology evolved 2020-2026
Ukraine proved something counterintuitive: dominance doesn't require superiority. A smaller, resource-constrained force used cheap drones to compete against a vastly larger military. But the victory was incomplete. Russia adapted. Both sides learned. By 2024, the tactical advantage had flattened.
This wasn't a drone victory. It was a discovery of a new kind of war.
Part one: warfare transformed
For decades, military advantage came from centralisation. Command flew from headquarters to field units. Intelligence flowed upward; orders flowed down. The side with the best communications and the fastest decision cycle won.
Ukraine broke this pattern. Centralised command became a liability. When your communications network is under cyber attack, when your command nodes are targeted by precision strikes, when an adversary controls the electromagnetic spectrum, a distributed force survives.
The shift happened in three waves.
Wave 1: Drones replace manned aviation (2022-2023)
In 2022, Ukraine deployed cheap FPV drones—off-the-shelf quadcopters modified for kamikaze attacks. A single drone cost $300 to $500. An armoured vehicle cost $1 million. The maths was brutal.
Russia deployed traditional air power. Manned helicopters and fixed-wing aircraft. They died in the air. Too vulnerable. Too expensive to lose. By late 2022, manned aviation was effectively withdrawn from active combat zones.
The Tactical Implication: Air superiority meant drone superiority. Fighters and helicopters became strategic assets too valuable to risk. Drones became tactical weapons.
Wave 2: Real-time ISR becomes the bottleneck (2023-2024)
By 2023, the constraint shifted. It wasn't firepower. It was information.
A drone could strike a target. But only if someone detected that target. And only if the targeting information reached the operator before the target moved. The time from detection to engagement compressed from hours to minutes.
This created a new strategic problem: persistent surveillance. Ukraine needed constant eyes on Russian positions. Not satellite passes that came twice a day. Constant. Real-time. Everywhere.
The solution was drone ISR networked with ground-based radar and satellite imagery. Multiple sensor streams fed into a single operational picture. Something detected by one system could be engaged by another, kilometres away. The OODA loop—observe, orient, decide, act—compressed to seconds.
This required more than hardware. It required the ability to fuse data from incompatible systems in real time. To trust machine recommendations without human review. To authorise strikes on algorithmic confidence scores.
Wave 3: Distributed operations & autonomous logistics (2024-2025)
By 2024, the third shift emerged. Centralised command was no longer sufficient. Ukrainian units in forward positions, cut off from reliable communications, needed to make decisions autonomously.
Supply lines were the critical vulnerability. Traditional logistics required trucks, roads, and rear bases. A truck on a road is easy to target. So Ukraine tested autonomous drone swarms for resupply. Small UAVs carrying ammunition, medical supplies, food. Coordinating without a central controller. Flying through contested airspace.
This required genuine autonomy. Not remote control. Not pre-programmed routes. Swarms that adapted in real time. Platforms that made decisions locally when comms were jammed.
The Tactical Lesson: Warfare required three simultaneous capabilities: real-time information fusion, distributed autonomous decision-making, and resilience to network degradation.
Part Two: Two competing philosophies
The defence technology industry split into two camps responding to these lessons. They represent fundamentally different bets on the future of warfare.
Camp 1: Palantir Gotham — fusion over autonomy
Palantir's approach is information dominance. Gotham is an ISR fusion platform. It ingests real-time data from every available source: drone feeds, satellite imagery, radar, signals intelligence, human reports. It fuses the streams in real time. And it outputs recommendations to human decision-makers.
The architecture is classical software systems: data layer, processing pipeline, user interface. Gotham handles the hard problem of reconciling incompatible data streams. Multiple sensors detect the same target at different coordinates. Gotham resolves the conflict. A radar reports movement at location A; a drone sees it at location B. Gotham weights the confidence of each sensor, updates the target track, predicts where it will be next.
The human stays in the loop. Gotham recommends a target. An analyst reviews. A commander approves. The firing solution is computed automatically.
Palantir built this system for the world where information is the constraint. Where human decision-makers are the bottleneck. Where trust in the machine is conditional on transparency and human judgment. The system is built on the assumption that warfare remains fundamentally under human control.
Revenue reflects this bet. Palantir generated roughly $4.48 billion in total global revenue in 2025. While commercial revenue represents a rapidly growing share of that total, its government and defence contracts remain massive foundational drivers of its business. Palantir is the dominant ISR platform in the US military.
Camp 2: Anduril Lattice — autonomy over fusion
Anduril's bet is different. Lattice is an operating system for autonomous systems. It assumes the human will be out of the loop.
The architecture is distributed. Autonomous platforms—drones, ground vehicles, ships—run local decision engines. These engines are trained using reinforcement learning on tactical scenarios. When a Lattice platform detects a target, it doesn't send information upward to a central fusion engine. It makes a local decision.
The key difference is latency. Palantir's architecture centralises fusion, which guarantees consistency but introduces delay. A platform makes a detection, sends data to cloud, waits for Gotham to process, waits for a recommendation, waits for human approval. Fast by traditional standards. Still seconds. Sometimes longer.
Lattice's architecture distributes decision-making. A platform detects a threat, computes a response locally, acts. Milliseconds. No waiting for human approval. No cloud round trip.
But Lattice can't work alone. It requires coordination. Multiple autonomous platforms need to agree on tactics. Lattice handles this with decentralised consensus algorithms. Each platform shares its local state with neighbours via mesh networking. No central command required.
The tactical assumption is stark: humans will authorise drone swarms to operate autonomously in contested environments. To make tactical decisions without explicit approval for each action.
Anduril is the bet that warfare will move faster than humans can follow. That distributed autonomy is not optional but necessary. Anduril is private, and as of early 2026, its valuation has skyrocketed to $60 billion. It is no longer just a fast-growing start-up, but a disruptive force rivalling legacy primes. The company spent 2024 and 2025 securing monumental government programmes—including the USAF's Collaborative Combat Aircraft (CCA) programme and a historic 10-year, $20 billion enterprise software contract with the US Army to deploy Lattice at an unprecedented scale.
The incompatibility problem
These philosophies are not compatible. A system designed around human-in-the-loop decision-making optimises for transparency and slow, deliberate choices. A system designed for autonomous execution optimises for speed and local optimisation. The trust models are opposite. The latency requirements are opposite. The user interfaces are opposite.
This creates a market bifurcation. Palantir dominates strategic ISR and intelligence analysis. Anduril is winning in autonomous platforms and swarms. They are building for different customers with different needs.
The US military will likely deploy both. Gotham for intelligence fusion at headquarters. Lattice for autonomous execution at the edge. But integration will be painful. The systems don't speak the same language.
Part Three: The OEM reckoning
The traditional defence contractors—Northrop Grumman, Raytheon, General Dynamics, Lockheed Martin—built their business on selling platforms. F-35 fighters, missile systems, radar suites. Hardware. Heavy. Expensive. Durable.
These companies are now facing the reality that Palantir and Anduril have already internalised: modern warfare is software. Hardware without software is inert.
The response from the OEMs has been acquisition and integration. Northrop partnered with Palantir to integrate Gotham into its ISR platforms. Raytheon is building "modular weapon systems" that can swap software modules. General Dynamics is moving towards "cloud-native architecture" for its combat systems.
But acquisition is expensive. Integration is slow. The talent is migrating to start-ups that move faster. A venture-backed autonomy company can ship a new feature in weeks. A traditional OEM requires months for approval, security review, and integration testing.
The venture funding data is unambiguous. In 2025, global defence tech venture funding was staggering, tracking towards roughly $50 billion annually (following nearly $38 billion in 2024). The bulk of it went to start-ups focused on autonomy, ISR/AI, logistics, and cyber. OEMs still dominate traditional hardware spending, but the explosive capital growth is entirely in the software layer.
The consolidation is inevitable. With Anduril already hitting a $60 billion valuation, legacy primes like Northrop and Raytheon have received a definitive signal: they are no longer just looking at nimble acquisition targets; they are facing formidable, capital-rich competitors.
The OEM that can successfully integrate software dominance into its existing business will survive. The one that can't will become a platform provider for others' software—essentially becoming a component supplier rather than a systems integrator.
Part Four: The real emerging threat
While the industry obsesses over drones and autonomy, the real strategic competition is happening in space.
Persistent surveillance from orbit
For decades, satellite ISR was a low-refresh problem. A satellite passed over a target once per day. By the time the imagery reached an analyst, it was 24 hours old. Strategic, not tactical.
That changed with LEO constellations. Companies like Planet Labs deployed hundreds of small satellites in low Earth orbit. They pass over the same location multiple times per day. Latency dropped from hours to minutes. Coverage became persistent.
By 2026, real-time satellite imagery is a commodity. Ukraine uses it. Ukraine gets it fast enough to be tactically useful.
The implication is that any future conflict will be visible from orbit in real time. A troop movement is detected. Positioning data is disseminated. An autonomous swarm is dispatched to engage.
This creates a new vulnerability. If your military operations can be seen from space in real time, they can be attacked in real time. Massed formations don't work. Logistical nodes become indefensible. The entire operational model shifts.
The counter to persistent satellite ISR is anti-satellite capability. Shoot down the satellites. China and Russia are developing kinetic anti-satellite weapons. The US is racing to defend its constellation.
Hypersonics & directed energy
Hypersonic missiles are the answer to traditional air defence. Travel at Mach 5 and above. Move too fast to intercept with existing systems. Russia used them operationally in Ukraine. China and the US are deploying them.
But hypersonics have a more subtle strategic purpose. They bypass the air defence network. They make layered defence architectures obsolete. A force that relies on missiles and interceptors is suddenly vulnerable to threats it can't stop.
Directed energy weapons—high-power lasers and microwave systems—address a different problem. Traditional missiles have minimum engagement ranges. Directed energy weapons don't. They also have infinite magazine capacity at the energy source. Fire one laser beam, then fire another. No reloading.
Both hypersonics and directed energy are in field trials or early deployment. Neither is yet decisive. But both point towards a future where traditional defence systems are obsolete.
Cyber-kinetic integration
The most dangerous development is the integration of cyber and kinetic operations into a single kill chain.
In Ukraine, Russian forces attempted to disable Ukrainian air defence systems by cyber attack before launching cruise missiles. In multiple cases, this worked. Electronic warfare disabled communications. Cyber attacks isolated systems. Kinetic strikes followed.
This is the doctrine that will dominate future warfare. Cyber as preparation. Kinetic as execution. Not separate domains but integrated operations.
The defender must assume that any system connected to the network can be compromised. Any sensor can be spoofed. Any communication can be jammed. The only resilient systems are those that can operate autonomously without network dependency.
This is the fundamental insight that drives Anduril's architecture. Lattice platforms assume they will be operating in a contested environment where the network is unreliable. Local autonomy is not optional. It's survival.
The strategic question
We have entered a period of military technological acceleration. The technologies that dominated warfare for 70 years—manned aviation, centralised command, layered air defence—are becoming obsolete. The technologies that will dominate for the next 30 years are still being defined.
Palantir is betting that information fusion and human-machine teaming will remain the centre of gravity. That humans will always be in critical decisions. That warfare will remain fundamentally deliberate and controlled.
Anduril is betting that speed will dominate. That distributed autonomy will beat centralised intelligence. That humans will authorise machines to operate faster than they can perceive.
One of these bets is wrong. Possibly both. The future will likely be messier than either company imagines. But the strategic implications are clear.
The nation that masters distributed autonomous systems operating in contested environments will dominate warfare for the next 30 years. The nation that can integrate real-time space-based ISR with autonomous execution will have overwhelming advantage.
The US has advantages in both categories. But China and Russia are accelerating. The consolidation in the US defence industry is happening. Venture funding is flowing. The OEMs are acquiring and integrating.
But speed matters more than size. The start-up that can move faster than the bureaucracy will shape the outcome.
The question isn't whether this technology will be adopted. It's who will adopt it first, and at scale. That answer will determine the strategic balance for a generation.