The Grid Is the Front
A $2,000 fiber-optic drone now defeats the concrete built to protect a $3.5 million transformer. Ukraine got through the worst winter of the war by decentralizing its networks; Europe, already under hybrid attack, meets none of the standards Ukraine set under fire.
Both sides of the war in Ukraine now treat civilian infrastructure as the decisive theater. A $2,000 fiber-optic drone defeats a concrete sarcophagus built to protect a $3.5 million transformer. Ukraine's answer is decentralization; Europe has neither the protection nor the topology.
This is the fourth post of our defense series. The first mapped the European defense-tech landscape and its capital structure. The second described the productive-forces shock: the drone has displaced the shell. The third traced the supply chain both camps share. This one follows the war to where it is now mostly being fought: the electrical substation, the compressor station, the pumping plant. And it asks what a European investor should conclude from the fact that this front moves faster than any other.
In May 2026, Russian crews operating near Sumy worked out a two-step routine. A first FPV drone, guided through a hair-thin fiber-optic cable and therefore immune to jamming, tears open the anti-drone netting stretched over an electrical substation. A second drone flies through the hole, or through the ventilation openings of the concrete sarcophagus underneath, and detonates against the autotransformer inside. The drone costs about $2,000. The 330 kV autotransformer costs about $3.5 million and takes months to replace. Reuters verified at least four such strikes on substations that Ukrainian engineers described as well defended.
The sarcophagi were themselves a response, designed in 2023 against Shahed drones, and they worked. The tactic that defeats them is roughly three months old. This is the tempo of the infrastructure war: every defensive answer has a shelf life, and the shelf life is now measured in weeks.
The report nobody wanted to write twice
In 1945, the United States Strategic Bombing Survey sent a thousand investigators into the rubble of Germany to establish what strategic bombing had actually achieved. Its Utilities Division delivered one of the most uncomfortable findings of the entire survey. The German electric grid, the report concluded, had been "in a precarious condition from the beginning of the war." German planners knew it: the secret minutes of the Central Planning Committee discussed at length their fear that the Allies would discover the grid's extreme vulnerability. The Allies never did. They had assumed the network was too redundant, too meshed, too repairable to be worth the sorties, and spent the tonnage on cities instead, where the same survey found production had kept rising until mid-1944. The decisive target of the industrial war went essentially unbombed because the attacker misjudged the topology of a network.
Eighty years later, Russia is running the campaign the Allies debated and rejected, against a grid it knows intimately because Soviet engineers built it. Around 5,500 drones and missiles per month were directed at Ukrainian energy targets during the past winter, in salvos of up to 700 munitions in a single night. And the campaign has answered the question the USSBS left open: a modern grid can be broken by sustained attack. Ukraine entered the war with 55 GW of generating capacity. By February 2026 it had about 11.5 GW available, in the coldest winter since 2010, with district heating plants deliberately targeted at minus 20°C. The UN's human rights monitors documented what a grid attack actually is in an urbanized country: when the substation dies, the water pumps stop, the mobile network goes dark within hours, the pipes freeze and burst, the sewage backs up. Seventy percent of Ukraine's urban population lives in apartment buildings; electricity is their single point of failure.
Ukraine runs its own infrastructure campaign, aimed at the asset Russia cannot decentralize: hydrocarbons. Ukrainian drones hit Russian refineries 194 times in the first half of 2026 and reached all eleven of the largest, including Omsk, 2,500 km from the border. In late March, strikes on the Baltic terminals of Ust-Luga and Primorsk idled at least 40% of Russia's oil export capacity for a week. Refinery runs fell to their lowest level since 2009. Both belligerents have concluded, independently, that the highest-return targets of this war are civilian networks behind the front line. Neither is going back.
Four layers, one scoreboard
Four years of this have produced something that did not exist in any military manual in 2021: a tested practice of defending infrastructure under sustained attack. We read it as four layers: passive protection, active defense, redundancy, and cyber.
The passive layer has the weakest record. Ukraine's three-tier program (gabions against debris, concrete sarcophagi against Shaheds, full anti-missile enclosures for 22 critical substations) stalled at the point where costs became prohibitive. The sarcophagi, some 69 substations' worth, do stop Shaheds. The third tier was never built: a prototype tested in late 2023 performed well, and the program then collapsed into unpaid contractors and bureaucratic deadlock, at a projected cost of tens of millions of dollars per substation across a hundred sites. And the fiber-optic FPV now defeats tier two for $2,000. Concrete loses the cost race, then loses the adaptation race.
The active layer is where startups inverted the economics. Russia launched more than 54,000 Shaheds in 2025, at $35,000 to $50,000 each, against Patriot interceptors at $4.2 million a shot. No Western stockpile survives that arithmetic. The Ukrainian answer is a stack built almost entirely outside the prime-contractor world: a nationwide acoustic detection network of 14,000 sensors at $400 to $1,000 apiece (total cost below $54 million, less than thirteen Patriot missiles), feeding targeting data to interceptor drones that cost $1,000 to $5,000 and are produced by more than twenty companies. In February 2026, interceptor drones accounted for over 70% of the Shaheds downed over Kyiv and its outskirts, according to Oleksandr Syrskyi, then commander-in-chief, who reported some 6,300 interceptor sorties and more than 1,500 Russian drones destroyed in that single month (Ukrinform; the figure is Ukrainian and not independently verified, but the cost analysis does not depend on it). Roughly 100,000 interceptors were built in 2025; capacity is planned to grow eightfold in 2026.
Then, in May 2026, Russia fielded the Geran-4, a jet-powered Shahed flying at 500 km/h, and made the propeller-driven interceptor fleet obsolete in a matter of weeks. The counter-generation (jet interceptors, one manufacturer targeting 50,000 units per month from August 2026) was ramping as we wrote this. This is consumer-electronics cadence applied to air defense, a full weapons generation per quarter. Any investor who underwrites a counter-drone company on the assumption that its current product is the asset has misread the sector. The asset is the iteration speed.
The real defense is topology
The layer that actually got Ukraine through the winter is redundancy. Having concluded that a centralized grid cannot be defended at acceptable cost, Ukraine is rebuilding as a decentralized one, and the numbers show this is policy, not improvisation. Some 1.5 GW of small distributed gas generation is planned for 2026 alone, as much as the four war years combined. Solar additions hit 1.5 GW in 2025. Around 3 GWh of grid batteries were installed in 2025, putting Ukraine among Europe's five fastest-growing storage markets in the middle of a war. DTEK and Fluence built 200 MW / 400 MWh of storage, deliberately dispersed across six sites to shrink the target, in six months. The same logic runs through every network. Telecoms: mandatory national roaming, 120,000 batteries on cell towers, and the world's first wartime direct-to-cell service, launched by Kyivstar and Starlink in November 2025 and reaching 3 million users in seven weeks. The state itself: government registries migrated to hyperscale clouds in 2022, which is why four years of strikes on data centers have never taken down the Ukrainian state (RUSI's assessment is that this single decision saved the government).
Perfect protection of a network node is unaffordable, and temporary. What defeats an infrastructure campaign is topology: enough distributed generation, storage, routing and fallback that no single strike, and no thousand strikes, produces a decisive outage. The USSBS said exactly this about why the Allies spared the German grid; they believed its topology made it unbreakable. They were wrong about that grid in 1944. Ukraine is making the same proposition true on purpose.
The counterpoint: decentralization is a logistics problem before it is a technology problem, and Ukraine fumbled parts of it. Emergency boiler plants financed by donors sat in warehouses through the winter of 2025-26 because nobody had organized grid connections. Machines are the easy half.
The open problems
Mapping the twelve months of moves and counter-moves gives something close to an investment map of what remains unsolved. These problems are posed today in Ukraine, but none of them is Ukrainian: every line below applies to Europe's own networks, which is the subject of the next section. We set the map out the way we set out the substitution layer in our third post: by where the engineering bottleneck sits. Founders, this is for you:
Two features of this table matter for a VC. Almost every line is dual-use in the most literal sense: the same acoustic sensor, microgrid controller or OT security stack sells to a utility in peacetime and to a ministry in crisis, which resolves the single-buyer problem. And the software lines are the emptiest. The hardware layers already have twenty Ukrainian manufacturers and a dozen European challengers; but where lies the orchestration layer, the software that makes thousands of distributed assets behave like a grid under fire ?
Europe, measured against its own exposure
The hybrid campaign against European infrastructure is already running. Eleven Baltic subsea cables were damaged in the fifteen months to January 2025, which is what triggered NATO's Baltic Sentry patrols. Drone incursions shut European airports repeatedly through 2025, and Russian drones over Poland in September 2025 triggered Eastern Sentry. The market has already priced this: DroneShield, the listed pure play in site protection, booked record European contracts with revenue up 276%.
Against the scenario of open conflict, the relevant Russian capability is demonstrated, not hypothetical: coordinated salvos of several hundred drones and missiles per night, sustained for months, against grid, gas and heating targets, with strike ranges that cover most of the continent. What Ukraine absorbed in the winter of 2025-26 is the live-fire specification of that threat.
European infrastructure meets none of the Ukrainian defensive standards. No EU substation sits under a sarcophagus; there is no acoustic detection network over any European country; there is no mass interceptor inventory anywhere in NATO Europe, and the European Drone Defence Initiative targets full capability at the end of 2028, a timeline into which Russia fit the invention of the sarcophagus bypass roughly eleven times over. Europe's grid redundancy was engineered for storms and market coupling, not for deliberate attack. And the deepest vulnerability is industrial: large power transformers, the components Russia has learned to kill for $2,000, carry lead times of 2.5 to 4 years in peacetime, with global demand from data centers already outrunning supply and no meaningful strategic reserve anywhere in Europe.
Meanwhile the knowledge is for sale. Lithuania bought Ukraine's acoustic detection network in 2025. The US is deploying its derivative on American bases and negotiating to buy Ukrainian interceptors. General Cherry, Ukraine's largest interceptor maker, is opening production in New Hampshire. The one belligerent-grade infrastructure-defense stack in the world is exporting; the question is whether Europe is buying at the speed of the threat or at the speed of its procurement system (hmm...).
The window, once more
Every post in this series has ended on a window. This one concerns a category that does not officially exist yet. Dealroom and the NATO Innovation Fund count $8.7 billion of European defense, security and resilience funding in 2025; nowhere in that data is "critical infrastructure resilience" broken out as a segment, because nobody has yet drawn the line around acoustic sensors, interceptor drones, microgrid software, OT security and transformer alternatives and called it a market.
The Strategic Bombing Survey's investigators concluded that the decisive network of the last industrial war went unattacked because the attacker misjudged its topology. The attacker of this war has not repeated the mistake, and there is no reason to believe the next one will. Ukraine redesigned its topology under fire, in four years, through the hardest winters its civilians have known since the Second World War. That is what adapting after the fact looks like: survivable for a nation that has no choice, unacceptable as a plan for anyone who does. Europe still has the choice.
In every precedent this series has cited, institutions adjusted only after the crisis. The infrastructure war will be no different, except in what is at stake: this time the underlying asset is the continent's ability to keep the lights on, both literally and seriously.
Previous posts: Who owns Europe's next arsenal? · After the shell · Shenzhen on both sides of the front
This article was written with the help of Claude (Anthropic).
To continue the discussion, share your feedback at francois@newfundcap.com.