Why the Next Phase of Autonomous Warfare Will Be Won or Lost Not by Weapons — But by the Invisible Infrastructure That Connects Them
A Ukrainian commander launches eight hundred autonomous drones — a coordinated swarm of air and ground systems programmed to suppress enemy air defenses, identify artillery positions, and exploit gaps in Russian lines. The operation depends on real-time coordination: sensors feeding targeting data to strike platforms, movement algorithms synchronizing advance rates, and machine learning systems adapting to Russian countermeasures.[1]
Eighteen minutes into the mission, Russian electronic warfare assets sever the swarm's tactical ground uplinks to Western cloud infrastructure. The swarm doesn't abort — it continues operating on preprogrammed instructions. But it can't adapt. Russian forces rapidly relocate their artillery and air defense systems. Ukrainian sensors detect the movement but can't retask strike drones without cloud connectivity. What should have been a precisely synchronized operation devolves into hundreds of individual platforms executing obsolete instructions against targets that have already moved.[1]
This scenario hasn't transpired yet. But every condition required to make it inevitable is already in place.
The war in Ukraine is often described in the language of weapons. Yet a less visible element will shape the next phase of the conflict just as decisively as any piece of military hardware: the infrastructure to create and harness computational power.
Everyone talks about AI-powered drone swarms as a software problem — computer vision for target identification, autonomy for navigation, coordination algorithms for synchronization. In practice, the first constraint is bandwidth. A Carnegie Mellon study found that a single HD drone video feed at 25 fps consumes approximately 10 Mbps.[2] A Starlink terminal provides 10-30 Mbps of uplink. One feed can saturate the entire pipe.
The arithmetic is merciless. Even pulling a handful of video feeds for target designation — combined with telemetry data, encryption overhead, and packet retransmission for coordinating hundreds of additional drones — creates an operational bottleneck that no amount of AI sophistication can overcome. You can build the world's smartest targeting algorithm, but if it can't reach the drone, the drone flies blind.
Starlink has been a strategic asset for Ukraine — its low-earth-orbit constellation is inherently difficult to jam at the orbital level. But the ground terminals are not. They remain highly vulnerable to localized Russian electronic warfare.[3] When Starlink suffered a global outage in July 2025, Ukraine's military communications were immediately impacted.[4] More dangerously, Starlink has become a single point of failure. If these tactical uplinks were effectively denied — whether by Russian terminal jamming, cyberattacks, or corporate policy shifts — Ukraine possesses no alternative capable of sustaining its current command tempo.
The operational logic is unforgiving: forces must either process data locally at the edge — transmitting only compressed targeting summaries upstream — or accept that cloud-dependent systems will fail when links degrade.
Ukraine's 2022 cloud migration was brilliant crisis management. Within months of the invasion, more than 10 petabytes of state data — from ministries, universities, private firms — had shifted to Western cloud infrastructure.[5] This preserved the Ukrainian state under fire. But state continuity is not warfighting.
The cloud migration solved one vulnerability (domestic servers exposed to Russian missile strikes) while creating another: total dependence on contested network pathways now central to warfighting capability. Russia has exploited this with surgical precision.
Russian attacks on Ukraine's power grid have destroyed approximately 9 GW of generating capacity — roughly half of prewar levels.[1] This creates a cascading failure: energy destruction reduces capacity for domestic compute infrastructure, which increases dependence on external cloud services accessed via networks that Russia can interdict. The same threats that create the need for computation are destroying the infrastructure required to sustain it.
Russia, meanwhile, pursues computational autarky — accepting lower performance in exchange for sovereign control. Moscow is deepening AI cooperation with China, investing in domestic data-center capacity, and announced a 200% increase in military spending for 2025-2026 with significant allocations to domestic technology infrastructure.[6] Western cloud is technically superior. But Russia's approach is building resilience through systems that can't be severed by adversary action.
What Ukraine needs — and what any military force will need for autonomous operations at scale — is a layered computational architecture that doesn't collapse when any single layer goes dark.[1]
On March 13, 2026, L3Harris and Shield AI announced a successful demonstration of autonomous electronic warfare that points toward the compute-native battlefield. Their DiSCO (Distributed Spectrum Collaboration and Operations) system combined L3Harris's spectrum battle management with Shield AI's Hivemind autonomy software to create something new: drone swarms that autonomously manage the electromagnetic spectrum without human intervention.[8]
In a hardware-in-the-loop simulation, multiple unmanned aerial vehicles continuously monitored electronic signals, created a unified view of the electromagnetic spectrum, and adapted their behavior as conditions changed. AI algorithms determined how each platform should react to detected signals — locating hostile radar emissions, monitoring communications, supporting countermeasures — all at machine speed.[9]
This is the compute war made physical. DiSCO doesn't just fight in the spectrum — it thinks in the spectrum. Each drone is simultaneously a sensor, a processor, and a weapon. The electromagnetic environment becomes both the battlefield and the computing medium. The system validates that distributed edge intelligence can operate where centralized cloud cannot.
Lauren Barnes, L3Harris's President of Spectrum Superiority, stated: "By integrating autonomous decision-making with advanced battle management technology, we're answering the Pentagon's urgent call for coordinated command and control of multiple unmanned systems."[8]
Ukraine's ground robot experience reinforces the compute lesson from below. The Modern War Institute at West Point reports that Ukrainian forces now control UGVs using four different communication methods — Starlink satellite links, radio frequency line-of-sight, fiber-optic cables, and aerial relay nodes — depending on range, terrain, and electronic warfare conditions.[10] Despite these options, operators still struggle to maintain reliable links due to intense Russian jamming.
The numbers tell the story of scale. In 2025, Ukrainian industry delivered 15,000 unmanned ground vehicles to frontline units — up from 2,000 in 2024.[10] The 3rd Assault Brigade's UGV company conducts 80% of logistics operations with robots, reducing human exposure. In contested cities like Pokrovsk and Myrnograd, that figure reaches 90%.[10] Ukrainian General Staff reports robotic platforms have reduced personnel casualties by up to 30%.[10]
But every one of these platforms requires connectivity to operate — and every connection is a vulnerability. The lesson emerging from four years of combat: the robot that can think for itself when the link drops is worth ten that can't. Edge compute isn't a technical preference. It's a survival requirement.
Current Western aid to Ukraine has focused overwhelmingly on kinetic systems — artillery, air defense, armored vehicles. Very little has gone toward computational infrastructure.[1] This made sense when the primary challenge was state survival. As the war evolves toward autonomous operations at industrial scale, aid priorities need to evolve accordingly.
The compute war reveals a truth that applies far beyond Ukraine: the limiting factor in autonomous warfare is not the weapon — it's the wire. Every drone, every robot, every AI targeting system depends on an invisible infrastructure of bandwidth, processing power, and connectivity. Destroy the compute layer and the smartest weapons become dumb munitions.
This connects directly to the broader Brownian Report network. Ghost Spectrum showed us that AI can learn to jam signals it has never encountered. GPS-Denied Nav showed us that platforms must navigate without satellites. Volt Typhoon showed us that China pre-positions inside critical infrastructure. The compute war is where all of these threads converge: the battlefield is migrating from physical terrain to computational terrain, and whoever controls the compute layer controls the fight.
Russia understands this. Its electronic warfare doctrine doesn't just deny GPS — it denies the entire connectivity fabric that AI-powered warfare requires. Every time Russian EW assets sever a Starlink uplink, they're not just jamming a signal. They're severing a drone swarm's brain from its body.