For the first time, the Kremlin has put a price on Ukrainian strikes against oil refining. The figure of 2.3 trillion rubles looks alarming, but the real bill, with its burned-out warehouses, idled trucks and Yandex's blacked-out cloud, is still being written.
On October 1, 2026, answering questions after his address to the plenary session of the Valdai Club, Vladimir Putin did something the Russian leadership had avoided since February 2022. He named the price. "They began striking refineries and openly declared their goal: to damage our economy. They achieved that goal in part, because by our estimates we have lost roughly one percent of GDP," the Russian president said.
One percent sounds like a rounding error. With nominal GDP, which the Finance Ministry projects at 232 trillion rubles for 2026, that comes to roughly 2.32 trillion. The sum is more than half of the liquid portion of the National Wealth Fund and almost a third of the record budget deficit. Putin specified neither the period nor the methodology.
A week later, Ukrainian drones knocked out three Yandex data centers in four days. On the evening the last of them was burning, U.S. President Donald Trump announced an "energy truce," which, judging by Kyiv's reaction, the parties learned about from his post. The war in the rear has finally moved from the genre of military dispatches into the genre of accounting, and in that accounting Moscow now has line items that appeared in no budget law.
The Valdai Admission: What Putin Said and What He Left Unsaid
Let us start with the budget, because that is where all the lines converge. On September 28, the Finance Ministry raised its estimate of the 2026 federal budget deficit to 7.3 trillion rubles, or 3.2 percent of GDP. The law passed in the fall of 2025 had projected 3.8 trillion and 1.6 percent. The gap is twofold. Anton Siluanov's ministry intends to cover it in two ways: by spending 459 billion rubles from the National Wealth Fund, about 11 percent of its liquid portion, and by increasing net borrowing by 1 trillion rubles, to 5 trillion. For comparison, in 2021 the fund's liquid portion reached a record 8.66 trillion rubles; by September 1, 2026, only about 4 trillion remained.
Against this backdrop, Putin's 2.32 trillion rubles no longer look like a margin of error but like a full-fledged expenditure line that no one ever approved. GDP growth in the first half of the year was a mere 0.6 percent. The lost percentage point, if Putin is to be believed, exceeds the economy's entire growth for the half year.
Let us break the figure down. According to available data, there were about 50 successful raids on Russian refineries in the first half of 2026, and roughly as many again from July through September. August was the war's record month, with at least 22 strikes on oil refining infrastructure. If Putin's percentage is attributed to the current year, the average successful strike costs the Russian economy about 23 billion rubles. The estimate is rough, built on an interpretation of the Kremlin's words, but it establishes the order of magnitude.
Why did Putin need to name a figure at all? The answer lies in the same remark. The president framed it as "strikes on civilian targets" and immediately declared that, after Russia's retaliatory strikes, Ukraine "now has no metallurgical industry." This is the classic symmetry of damage: acknowledge your own losses in order to justify the ones you inflict. The same day, Putin called "absurd" the proposals to trade a halt to Russian strikes on Ukrainian ships for a halt to Ukrainian strikes on refineries. He spelled out the logic plainly: even if the attacks stopped, sanctions would not allow Russia to increase diesel exports, while pressure on Ukrainian shipping hurts Kyiv.
I would read this remark more carefully than the audience in the hall did. Moscow signaled that it values its leverage against Ukrainian ports more than a repair window for its own plants. That kind of trade-off is made by a side that believes it can outlast its adversary.
19 Billion Rubles per Unit: Why a Russian Refinery Cannot Simply Be Fixed
Oil refining occupies a paradoxical place in Russia's fiscal system. The bulk of budget revenue comes from crude oil, not gasoline. Refineries generate non-oil-and-gas revenue, and through the damper mechanism and the reverse excise tax they sometimes even draw federal money themselves. When a refinery goes down, the crude goes to export, the budget may even save on the reverse excise in the short run, and the main blow falls on the business owner, on fuel consumers and on the entire chain of intermediaries. That is why the real cost of the strikes shows up in repairs, downtime and gas stations, not in Treasury reports.
The Energy Ministry does not publish what repairs cost. Indirectly, however, it has disclosed its arithmetic. From 2011 to 2018, Russia reconstructed and commissioned 95 secondary refining units, with investment totaling 1.6 trillion rubles, or less than 17 billion per unit. From 2019 through 2026, only 27 of the 40 planned units were commissioned, at a cost of 512 billion rubles. That works out to almost 19 billion per unit. The figures for the two periods corroborate each other and capture the rising costs, driven in part by sanctions.
Anton Rubtsov, director of the Energy Ministry's oil and gas department, once acknowledged that foreign technologies dominated the original modernization projects. The Ministry of Industry and Trade tried to organize import substitution, but in his assessment the industry still gravitates toward Asian suppliers. Translated from bureaucratese: Russian oil refining was built on Western equipment, and now it has to be repaired with Chinese equipment.
There are corporate benchmarks as well. In 2023, Gazprom Neft launched a new primary refining complex in Omsk costing more than 66 billion rubles. It replaced six units of the previous generation, meaning each replacement cost about 11 billion. Back in 2024, Western industry analysts estimated the damage from a single drone attack at anywhere from several million to several tens of millions of dollars. At that time the strikes were pinpoint. In 2026, drones have learned to target specific units, and after repeated attacks the plants require major overhauls.
A major overhaul takes at least six months. The unit has to be designed at a specialized institute, approved, manufactured and installed. After a fire, add inspection, removal of charred structures and cleanup. With imported equipment, timelines stretch even further, and some items are simply unavailable.
Kapotnya Burns on Election Day: Anatomy of the Country's Most Expensive Repair
Gazprom Neft's Moscow refinery in Kapotnya is the symbol of the entire campaign. The plant, with a capacity of about 11 million tons a year, sits within the capital's city limits, next to residential neighborhoods. By industry estimates, it supplies about 40 percent of the fuel consumed in Moscow and roughly 70 percent of the gasoline and jet fuel in the Moscow region.
On June 16, 2026, a strike disabled the AVT-6 primary distillation unit, which accounted for about 53 percent of the plant's capacity. On June 18, drones hit the Euro+ combined unit, which handled the remaining 47 percent. The plant lost all of its capacity, and lines formed at gas stations in the capital and the surrounding region. On the night of September 20 came at least the fourth attack since mid-May.
The date was not chosen at random. September 20 marked the end of three days of voting in the State Duma elections. Russia's Defense Ministry reported that 1,110 Ukrainian drones had been shot down overnight over the country's regions, Crimea and the Black Sea, and Moscow Region Governor Andrei Vorobyov said 249 of them had been downed over 17 of the region's districts. The strike also hit Combined Heat and Power Plant No. 22, which supplies the refinery with process steam. Primary refining was halted completely, and starting September 21, the refinery's products were no longer offered for trading on the St. Petersburg International Mercantile Exchange.
The damaged Euro+ complex was launched in 2020 as a showcase of modernization. Investment totaled 98 billion rubles, and its capacity is 6 million tons of crude a year. The complex comprises five process blocks: the ELOU-AVT-6 primary refining unit, gasoline reforming, diesel hydrotreating with an isodewaxing block, gas fractionation and amine regeneration.
An ELOU-AVT-6 cannot be bought. Facilities like this are not sold off the shelf; they are built for a specific plant to a custom design. Publicly available benchmarks are scattered across years and countries. In 2012, a new ELOU-AVT-6 in Russia was valued at 8.3 billion rubles. In 2017, contracts to build units with one-third of that capacity were valued at 14 billion.
In Baku, this arithmetic is familiar firsthand. Back in 2018, dismantling and installing an operating ELOU-AVT unit at the Heydar Aliyev Baku Oil Refinery, a unit one-third the size of Moscow's, cost 75 to 80 million dollars, or 6 to 7 billion rubles at the exchange rate of the time. The difference is that Azerbaijan carried out its modernization with open access to the global equipment market, without sanctions lists or shady intermediaries. Moscow is denied that luxury.
The China Multiplier of Four: The Sanctions Markup the Energy Ministry Won't Talk About
"The Russian factor relative to China is four." That is how one market participant explained the rising cost of repairs. Chinese equipment, once delivered to Russia, costs four times its list price. European components are routed through China only in exceptional cases, at prices the industry describes as sky-high.
What lies behind this multiplier can be seen in actual projects by Chinese contractors. In 2016 and 2017, China Petroleum Engineering and Construction Corporation upgraded the atmospheric distillation at an Algerian refinery for 408 million dollars under an engineering, procurement and construction contract that included naphtha hydrotreating and a new control room. At Kuwait's Al-Zour refinery, the main units, including atmospheric-vacuum distillation, were built by Sinopec Engineering in a consortium with Spain's Técnicas Reunidas and South Korea's Hanwha, under a turnkey contract for 2015 through 2019 worth 4.1 billion dollars, about 265 billion rubles. China's Zhongke refinery, with a capacity of 10 million tons a year and built between 2017 and 2020, cost about 5.6 billion dollars in its first phase, or around 360 billion rubles.
Kapotnya brings together the same processes: atmospheric-vacuum distillation, diesel and naphtha hydrotreating, hydrocracking and sulfur recovery from hydrogen sulfide. Take Chinese prices and multiply them by the Russian factor of three to four, accounting for logistics, customs, adaptation to Russian standards and intermediaries' margins, and the lower bound for restoring the Moscow refinery comes to no less than 25 billion rubles. The upper bound approaches half a trillion. The spread is enormous, but both figures explain why Gazprom Neft is in no hurry to publish its own cost estimate.
A precedent is worth recalling. On September 14, 2019, a drone and cruise missile attack on Saudi Aramco facilities at Abqaiq and Khurais knocked out about 5.7 million barrels per day, roughly half of the kingdom's output. By the end of September, the company announced that production had been restored. The Saudis made repairs with open access to every supplier on the planet and a bottomless wallet. Russia is making repairs through intermediaries, at quadruple rates and under fresh strikes that arrive before the previous repair is finished.
The Statistics Have Gone Dark, but the Trucks Don't Lie
The consequences for the fuel market cannot be assessed directly. Since May 29, 2024, Rosstat has not published data on gasoline and diesel production. By shutting down the statistics, the state has condemned everyone, its own agencies included, to guesswork.
And the guesses diverge. In early July, the General Staff of the Armed Forces of Ukraine spoke of 42.74 percent of Russian capacity knocked out. Western estimates at the end of July distinguished between nominally disabled capacity, about 45 percent, and capacity actually standing idle, more than 30 percent. Alexander Shokhin, head of the Russian Union of Industrialists and Entrepreneurs, put the losses at 25 to 30 percent. The industry consensus puts refining throughput at 3.8 to 4.2 million barrels per day, compared with 5.3 to 5.5 million in 2024 and 2025, a decline of 20 to 25 percent. In September, the International Energy Agency lowered its baseline forecast for the next 18 months to about 4 million barrels per day. The Ukrainian figure is, of course, inflated for propaganda purposes, while Shokhin's Russian estimate errs on the side of caution, but both agree on the main point: the country has lost between a quarter and a third of its refining.
Lost refining does not yet equal lost consumption. Part of the shortfall is covered by stockpiles, imports, rerouted flows and heavier loading of the surviving plants. In July, Putin acknowledged the shortage, calling it noncritical, and ordered a faster ramp-up of petroleum product output by small and medium-sized businesses. The order speaks volumes: when the largest plants are burning, the authorities suddenly remember mini-refineries.
Anyone who, like me, lived through the late Soviet economy will recognize this reflex. When statistics become inconvenient, they are shut down, and reality starts to be measured by indirect signs: the length of lines, empty shelves, the talk among drivers.
One independent model gauges the depth of the crisis through traffic accident statistics, which track traffic volume. In July 2026, the number of accidents came in 7.9 percent below the expected level; preliminary August data put it 12.4 percent below. The deviation is especially pronounced outside populated areas, that is, on highways. Allowing for later revisions, the model's author estimated the August decline in gasoline consumption at 7.5 to 10 percent.
The second indicator is more reliable. The Platon system, owned by a company belonging to Igor Rotenberg, charges per-kilometer fees on trucks weighing more than 12 tons on federal highways. It has operated since November 15, 2015, and reports as of the 15th of each month. Monthly payments can be reconstructed from the operator's reports by adjusting for the rate: 3.34 rubles per kilometer in 2025 and 5.19 rubles in 2026. The calculation shows that from August 15 to September 15, 2026, heavy trucks paid for about 1.93 billion notional kilometers, compared with 2.4 billion a year earlier. That is a drop of roughly 20 percent.
One month is not a trend, and I would not build far-reaching conclusions on it. Even if the real decline is half as large, and every tenth truck has stopped rather than every fifth, for a country spanning nine time zones and dependent on road logistics this is a tangible blow to everything that moves, is produced and is sold. Prices tell the same story. According to Rosstat, gasoline was 16.5 percent more expensive in January through August than in the same period of 2025, and in August alone prices were up 25 percent year over year.
444,000 Rubles per Square Meter: How the Online Marketplace Economy Is Burning
Retail trade generated 61 trillion rubles in turnover for Russia in 2025. It can hardly be called the core of the economy, but socially it is as sensitive as gasoline at the pump. Alexander Maslennikov, Deputy Secretary of the Security Council, described the effect of the strikes on Wildberries warehouses as "very limited" and "symbolic in nature," while acknowledging local losses and damage to small businesses.
Let us look at the symbolism in rubles. After the attacks on warehouses in Elektrostal, with 250,000 square meters, and Kotovsk, with 108,000 square meters, market participants estimated the cost of rebuilding the logistics complexes at 20 billion to 35.8 billion rubles. The spread reflects construction costs ranging from 60,000 to 100,000 rubles per square meter. The midpoint, 80,000, serves as a working benchmark for future attacks.
Burned inventory is harder to calculate, but it can be done. In Elektrostal and Kotovsk, it was valued at 150 to 170 billion rubles across 360,000 square meters, or about 444,000 rubles per meter. After the attacks on warehouses in Krasnodar, with 150,000 square meters, and Nevinnomyssk, with 100,000 square meters, inventory losses were estimated at 100 to 120 billion, or roughly 440,000 per meter. Two independent pairs of warehouses produced nearly identical figures, which says more about their reliability than any Security Council statement.
The main damage does not burn. Every day a distribution center sits idle deprives tens of thousands of sellers of revenue, releasing goods after an attack takes weeks, and loan interest keeps accruing. A seller who loses his stock loses his working capital: to get back to trading, he has to buy goods all over again. Many have no reserves, and they go off the books or shut down.
The courts are already recording this. According to the arbitration court database, from July 15 to September 15, 2026, Wildberries was named as defendant in 599 new lawsuits, 53.6 percent more than a year earlier. Ozon's main legal entity accumulated 236 new cases, more than double the previous year's total.
For the Finance Ministry, this means direct losses to the tax base. If sellers' lost revenue amounts to 445 to 508 billion rubles, one-time budget losses fall in the range of 13 to 30 billion. The lower bound is corporate income tax at a 15 percent margin; the upper bound is the 6 percent simplified-system tax levied on all revenue. Far grimmer figures are circulating in the market: supposedly, Wildberries' turnover has shrunk by a quarter as sellers have left. With the company's annual turnover at 6.1 trillion rubles, that would mean 1.5 trillion in lost revenue and about 330 billion in value-added tax alone, at the 22 percent rate in effect since January 1, 2026. I am skeptical of this estimate: it is not confirmed by financial statements and most likely conflates the effect of the attacks with a general cooling of demand. Even the academic estimate by the Center for Tax Policy and the Institute of Economic Forecasting of the Russian Academy of Sciences holds that, because of the way online marketplaces operate, the budget may collect 440 to 690 billion rubles less from them in 2026 than expected. The attacks merely accelerate this trend.
The long-term effect is more dangerous than the one-time hit. The departure of even 10 percent of active sellers, who generate roughly 610 billion rubles in turnover, would mean a tax shortfall of 35 to 50 billion rubles a year, plus another 7 to 10 billion in social insurance contributions. If 5 percent leave, the losses are halved; if 20 percent leave, they double.
Meanwhile, inflation is gathering speed. After the attacks, prices on online marketplaces rose 10 to 30 percent within a few days. In the first half of 2026, insurance payouts to businesses grew by a third. Loss ratios in terrorism insurance exceeded 100 percent; insurers are cutting liability limits and declining to cover some risks altogether. Energogarant estimates that rates for terrorism and sabotage insurance have risen tenfold over four years. Sooner or later, every ruble of insurance premium ends up on the price tag. Businesses, meanwhile, are shelving investment plans, and the logic is clear: a new workshop can go up in flames from a single drone costing a few tens of thousands of dollars.
The Week the Cloud Fell
On the night of October 8, drones attacked a Yandex data center in Sasovo, in the Ryazan region. The facility, built in 2012 and brought online in 2014, sits on the grounds of the Sasta machine-tool plant, which the Ukrainian side links to the defense industry. The fire affected part of the infrastructure, and operations were shut down completely. Yandex shares on the Moscow Exchange fell more than 3 percent that morning. On October 9, drones damaged several modules of the data center in Kaluga, in the Grabtsevo industrial park, which is rated at 63 megawatts and considered the company's most powerful facility. On October 10, Yandex published a new version of its user agreements, adding drone attacks to the list of force majeure circumstances. A few hours later, on the night of October 11, the data center in the Energetik neighborhood of Vladimir caught fire.
The force majeure detail is telling. The country's largest technology company legally acknowledged that drone strikes had become a routine business risk before the state acknowledged as much politically.
The three attacked centers housed all three of the company's supercomputers, with a combined performance of 50.36 petaflops. Lyapunov and the most powerful of them, Chervonenkis, were in Sasovo; Galushkin was in Vladimir. "At this time, we cannot confirm that the equipment at the Sasovo data center can be restored," Yandex said. In another statement, the company admitted it was still very difficult to give a timeline for recovery. There is no reliable confirmation that the supercomputers were destroyed, and no proof to the contrary. If they were damaged, the blow will fall on the development of YandexGPT, which these machines are used to train.
The supercomputers ran on chips from America's Nvidia, whose shipments to Russia are banned under export controls. That means restoration will go through gray schemes, with the same China multiplier and timelines of up to six months. Building the Sasovo center cost 6 billion rubles in 2011, and its reconstruction a decade later another 700 million. Investment in the Vladimir center was estimated at about 9 billion. The company does not disclose the value of the computing capacity itself, and the scale of the losses has not been officially assessed.
More important than the hardware is the cascade. After the October 11 strike, the Yandex Cloud platform went into emergency mode because two availability zones, ru-central1-a and ru-central1-b, were down at the same time, and the company described its remaining resource configuration as unstable. The outages reportedly affected at least six countries. Starting at 6:00 a.m. Moscow time, Rosselkhoznadzor, the federal veterinary and phytosanitary watchdog, switched two federal information systems, VetIS and Saturn, which handle veterinary control and pesticide traceability, to emergency mode. Disruptions were reported at T-Bank, Russian Railways, ticketing services, streaming platforms, online education platforms, Magnit and Fix Price. Logistics companies complained that the electronic waybill, which had only just been made mandatory, had stopped working. "A couple more strikes like that, and we'll be back to paper," one market participant said. Yandex hastily arranged with Selectel, K2Cloud and VK Cloud to migrate affected clients' data.
Herein lies the week's main strategic lesson. The digitalization of the Russian state in recent years was built on the principle of concentration: government systems, banks, logistics and retail converged on a handful of providers because that was faster and cheaper. In peacetime, concentration meant efficiency. The drone war has turned it into a target map.
A Truce via Truth Social: Who Really Benefits
An energy truce has been on the table for more than a year. In March 2025, after a conversation between Trump and Putin, the parties agreed to a 30-day moratorium on strikes against energy infrastructure, spent the entire month accusing each other of violations, and in April the Kremlin declared the arrangement exhausted. The Americans proposed a truce again in February 2026 at talks in Abu Dhabi. In late June, by Putin's own account, he rejected Kyiv's offer to stop strikes deep inside Russia in exchange for a reciprocal step. On September 22, on the sidelines of the 81st session of the UN General Assembly, Trump and Volodymyr Zelensky discussed the matter.
Trump made no secret of Washington's motive. He confirmed that he had asked Zelensky not to strike Russian refineries because the strikes were creating diesel problems for American farmers. He blamed the rise in global diesel prices primarily on the war in Ukraine rather than on the Iran crisis. The U.S. midterm elections will be held on November 3, 2026, and the price of diesel in farm states matters more to the White House than any percentage of Russian GDP.
On the morning of October 11, Zelensky said he was prepared to stop striking refineries if Russia halted its attacks on Ukraine's energy sector. That evening, Trump wrote that the truce was taking effect immediately, adding: "Don't violate it!" Zelensky acknowledged to American reporters that he had learned of the deal from that post: "If President Trump has agreed on this with Putin, please let us know when the truce begins." The Kremlin, through Dmitry Peskov, called the idea a very good one but immediately added that the safe operation of refineries would not solve global problems. What matters more to Moscow, by its own account, is the lifting of sanctions on supplies and the security of its tankers.
The distribution of benefits is plain to see. Russia gets a repair window without formal concessions and tries to sell it as a goodwill gesture in exchange for sanctions relief. Ukraine risks losing the war's most effective economic lever, and at the very moment it has begun working at full strength. Washington gets cheap diesel ahead of the elections. The loser will be whoever stops first and then discovers that the adversary has merely regrouped.
History knows how campaigns against an enemy's fuel supply end. On May 12, 1944, American bombers began systematic strikes on German synthetic fuel plants at Leuna and other centers. Albert Speer later wrote in his memoirs that on that day the technological war was lost. Speer's repair crews restored the plants within weeks, but new raids came faster. Russia in 2026 is not Germany in 1944: it has China at its back, raw materials and a vast domestic market. Yet the mechanics are the same, and they are merciless to the defender: repairs cost more than the attack, and the attacker need only strike more often than the repairs can be completed.
Two Pillars Under Fire
The Russian economy has rested on two pillars. Commodity exports fed the budget, while domestic consumption generated demand and jobs. When one pillar sagged, the system was rescued by something else: expensive oil, fast-turnover retail or the digital services that gave the whole structure its speed. The strikes of 2026 hit oil refining, warehouses and the cloud simultaneously for the first time. Offsetting the collapse of one sector with another, as Russia managed in 2014, in 2020 and in the first years of the war, when either oil prices or a consumer boom propped up the statistics, is now simply impossible, because the strikes are landing on every sector at once.
Putin's percentage is most likely an understatement. It does not account for the marketplace warehouses, the sellers' lawsuits, the insurance rates or the downed cloud, and it was calculated before Yandex's October week. How the Kremlin intends to close the gap while the National Wealth Fund melts away and borrowing has already climbed to 5 trillion, neither the Finance Ministry nor the Central Bank has yet explained in any coherent way.
My forecasts can be checked. The Moscow refinery will not return to design capacity before March 31, 2027. The energy truce in the form Trump announced will not survive past November 15, 2026: by that date, the parties will be publicly accusing each other of violations, just as in the spring of 2025. The actual federal budget deficit for 2026 will exceed the Finance Ministry's September estimate of 7.3 trillion rubles. The ru-central1-b zone of Yandex Cloud will not return to its previous capacity before April 2027.
The most accurate indicator of this war will be found neither in Kapotnya nor in Sasovo. It will be lying in the cab of a tractor-trailer on the M-7 highway outside Vladimir: a paper waybill with a blue stamp that the driver is once again carrying with him, because the electronic one wouldn't load.