On 1 October 2026, motorists in Ulaanbaatar left their cars and marched toward parliament. The grievance was not abstract. Filling stations had again run to hours-long lines, waits reaching eight hours at the worst points, with some drivers sleeping in vehicles overnight and purchases capped, at stages of the shortage, at 50,000 tugrik a motorist. The Ministry of Industry and Mineral Resources (MIMR, Аж үйлдвэр, эрдэс баялгийн яам) responded with the only remedy immediately available to a state that does not yet have a functioning domestic fuel refinery: more negotiated cargoes from Russia, China and South Korea, including a Russian commitment of 25,000 tonnes of additional AI-92 and a Mongolian request for 71,000 tonnes of petroleum products a month from China. The queues were a domestic political event. The molecules behind them were not.

Mongolia consumes about 3 million tonnes of fuel a year, 97 percent of it imported from Russia, Prime Minister Nyam-Osoryn Uchral said on 4 September at the Eastern Economic Forum. Demand, he added, is expected to reach 4 million tonnes by 2028. A 2024 intergovernmental agreement provides for annual Russian supply of 1.8 to 1.9 million tonnes of petroleum products and 60,000 tonnes of jet fuel. Ministry figures put the Russian share at 95 percent at the end of 2025 and about 97 percent in April 2026, against 96.84 percent in 2024. The residual has come mainly from China. Fuel is also the country’s largest import category, on the order of one-third of the import bill. A mining economy that ships coal and copper south therefore runs, almost liter for liter, on refined product moving south from Russian plants, above all through the Ulaanbaatar Railway system.

The 2026 break in that chain was collateral, not a cutoff. Ukrainian drone strikes damaged Russian refineries through the summer. One market tally counted 29 strikes in August and 12 more in the first three weeks of September, with one consultancy estimating unplanned idle capacity at roughly half of standard primary refining capacity before accounting for plants operating above design capacity and mothballed units returning to service. Russia imposed successive export restrictions during the summer, including a diesel restriction announced on 8 July and a broader fuel-export ban from 1 August through 31 January 2027. Intergovernmental deliveries, including those to Mongolia, were exempt, but exemption did not guarantee unchanged volumes. Shipments from the Angarsk refinery fell. Imports into Mongolia, which had been declining since April, were down 25 percent year on year in July, according to Bank of Mongolia (BoM; Монголбанк) data cited in reporting of the October protest. National stocks that officials had built to a 42-day cover between April and June dropped as low as six days. By late September they were back near 11 days — enough, the Fuel Emergency Task Force said, to keep distribution uneven and the queues in place. B. Dashpurev, state secretary at the MIMR, told the task force on 21 September that the disruption could last two to three years, the time he associated with repair of damaged Russian capacity. The reported wholesale import cost for AI-92 gasoline cited by the task force had increased, with the market premium adding a further $85 to $105 a tonne to the baseline contract rate over the preceding month. The shortage itself had already run about 50 days.

That chronology matters, because the strongest defense of the existing arrangement is a good one. Russia did not turn the tap. It kept the trade agreement in force, confirmed prior volumes with Energy Minister Sergei Tsivilev at the same Eastern Economic Forum, and supplied on preferential terms that a landlocked buyer with a narrow rail corridor has every reason to prize. Euro-2 AI-92 from established Russian contracts matches the fleet and the unloading system; Euro-5 cargoes from elsewhere do not drop onto the same logistics at the same price. Uchral used the forum to thank Moscow and to say that Mongolia would keep expanding fuel cooperation with Russia after its own refinery opens. For a state of 3.6 million people between two much larger neighbors, a discounted, rail-borne, contractually privileged supplier is not a failure of imagination. It is the rational baseline.

The baseline still failed the test that matters at the pump. A supplier concentration near 97 percent allows a refinery disruption in Russia to become a prolonged queue in Ulaanbaatar whether or not anyone in Moscow intends coercion. BoM data show the volume shock began months before the October march. Officials themselves now talk in years, not weeks. Price controls and a political commitment to hold down AI-92 — the grade used by most road vehicles — have kept the visible crisis from becoming a full price crisis, but they can encourage drivers to queue rather than absorb higher market prices, and they shift the cost of thin stocks onto the state. A previous shortage in October 2025, blamed then on Russian maintenance and logistics, had already produced multi-hour waits and provincial rationing, and had already prompted a cabinet order for a 30-day strategic reserve, which is generally maintained by private sector companies. The reserve was not there when the 2026 shock arrived. Recurrence is the evidence that this is a stock-and-concentration problem, not a one-off war story.

Diversification since the spring is real, and it is smaller than the announcements. On 30 July, after negotiations between MIMR Minister Damdinnyam Gongor and South Korea’s trade ministry, Seoul agreed to supply 50,000 tonnes a month: 10,000 of jet fuel, 20,000 of AI-92 and 20,000 of diesel. Set against current consumption, that headline contract is material.

A fifth of annual demand, if every tonne arrived. By late September the volumes that had actually moved were of a different order: 2,000 tonnes already imported from South Korea, with 12,000 tonnes described as ready, alongside an 8,000-tonne Chinese petrol cargo in transshipment at Zamiin-Uud and a preliminary 20,000 tonnes for the following month. The ministry has said it wants China’s share raised from about 5 percent toward 30 percent of consumption. Kazakhstan has been in talks since March; reports describe a Kazakhstan arrangement for at least 1 million tonnes a year, but the figure should be treated as an intended supply volume until cargoes actually move. Belarus has been approached for jet fuel. Customs duty and excise on imports were temporarily lifted. These are the correct instruments. They are not yet a second supply system.

Geography is the constraint the communiqués skip. South Korean fuel cannot reach Mongolia without transit through China or Russia, or a more expensive multimodal sea-and-rail route into a network built around Russian tank cars and the Ulaanbaatar Railway. China’s willingness to sell is also China’s control of the southern crossing. Raising Beijing’s share of fuel imports while China already takes more than 90 percent of Mongolian exports does not create an outside option. It stacks a second dependence on the neighbor that already prices the coal and copper which pay for the fuel. Kazakhstan is a more interesting hedge precisely because it is another producer with its own Russian-gauge rail problem and its own exposure to the same regional diesel tightness. A third supplier that cannot deliver through a third route is a contract, not a buffer.

The domestic refinery is the other number that has been asked to carry more political weight than its capacity can bear. The under-construction refinery at Altan Shiree in Dornogovi, financed by a 1.7 billion dollar Indian line of credit and executed with Engineers India Ltd, is designed for 1.5 million tonnes of crude a year, about 30,000 barrels a day. Construction began in 2018 against an original 2022 target. Indian officials said in October 2025 that operations should start in 2028; the Indian embassy reported the site 60 percent complete when Uchral visited in August 2026. Product plans cited by the government run to more than 300,000 tonnes of gasoline and 800,000 tonnes of diesel a year. Uchral’s own September formulation was that the refinery would meet 30 percent of total needs. A year earlier the industry minister had spoken of about half of domestic demand. Both can be defended, and the gap is the point. Nameplate capacity is half of today’s 3 million tonnes and three-eighths of the 4 million tonnes projected for 2028.

More importantly, the government's cited product slate—more than 300,000 tonnes of gasoline and about 800,000 tonnes of diesel—totals roughly 1.1 million tonnes of finished products, or only about 37 percent of today's three-million-tonne fuel requirement. Full run-rate is expected only years after start-up. The feedstock question remains unresolved: Mongolia produces crude and has exported it, and the industry minister has said talks with PetroChina are required to secure raw material. An Indian-financed refinery that depends on imported or politically allocated crude is a third-neighbor asset sitting on a two-neighbor supply chain. It will matter. It will reduce Russian dependence, but it will not eliminate it.

Storage is the cheaper half of the same problem, and the half that keeps being deferred. Officials have said queues ease only once cover reaches about 20 days; the country entered October with roughly 11. At 3 million tonnes a year, daily consumption is about 8,200 tonnes, so an 11-day national stock is on the order of 90,000 tonnes — a working inventory, not a strategic reserve. Construction of 300,000 cubic meters of new storage is underway. Even on a generous assumption that the tanks hold gasoline at roughly 0.74 tonnes per cubic meter, that is about 220,000 tonnes, or something near 27 days of current burn, and only once the tanks exist and are filled. Successive governments have promised three to six months of cover. The promise has outlived several cabinets. A 20-day operational target is the honest near-term number. Anything described as strategic has to survive the kind of multi-month Russian outage Dashpurev is now forecasting.

The transmission into the real economy is what makes the queue more than a motorist’s complaint. Mining remains central to the growth model. GDP rose 6.8 percent in 2025, to about $25.4 billion, supported by strong copper production at Oyu Tolgoi and continued coal exports to China. Diesel is the operating fluid of that model: haul trucks, generators at Gobi camps, the convoys running south to the border. Petrol rationing can quickly reach herders, because wells in the Gobi are pumped, not drawn by hand. Agriculture and the summer tourist season already lift demand sharply; reporting around the October protest put the underlying annual increase near 14 percent. When stocks fall, the first political pressure is urban and visible. The economic pressure is on the export machine that funds the budget. Fuel inflation also lands on the central bank. The BoM had already tightened, lifting the policy rate to 12.5 percent in August, with geopolitical fuel risk cited among the reasons uncertainty remained high.

This is where the third-neighbor idea either becomes a logistics program or remains a doctrine. The phrase entered Mongolian strategic thinking after U.S. Secretary of State James Baker offered the United States as a ‘third neighbor’ during his 1990 visit. It was written into the foreign-policy and national-security concepts as a multi-pillared balance to Russia and China: the United States, Japan, South Korea, India, European partners, later a wider set. It was never meant to replace the two borders. Mongolian governments have been consistent on that point, and the fuel crisis does not refute it. What the crisis measures is the distance between the doctrine and the physical system. India is financing the refinery. South Korea has signed the only third-neighbor supply contract with a monthly number attached. Neither has put 20 days of cover in a tank farm, nor a second rail path that does not run through a neighbor. The policy’s recent energy has been in minerals — uranium, critical-minerals diplomacy, coal and copper buyers — where third neighbors can write contracts without moving a barrel across the steppe. Fuel exposes the difference. An option that cannot arrive in a tank car is not an option on the day the Angarsk shipments slip.

Three paths follow from the arithmetic, and they turn on variables Mongolia only partly controls.

In the baseline, Russian refining recovers slowly, intergovernmental volumes resume near the contracted volumes, and Chinese and Korean cargoes fill the gap without ever reaching their headline shares. Stocks oscillate between one and three weeks. Queues return whenever a refinery, a border, or a tourist season moves the wrong way. The refinery starts late in the decade, ramps toward a fraction of demand, and imports from Russia continue, as Uchral has already said they will. Sovereignty in this path is managed dependence: real, cheaper than autarky, and still one drone campaign away from Sukhbaatar Square.

A more favorable path requires three things at once. The 300,000 cubic meters of storage is finished and filled to the 20-day line before the next winter. The Korean contract moves in the tens of thousands of tonnes a month rather than the low thousands, which means transit arrangements with China or Russia that survive a political argument. The refinery’s crude problem is settled early enough that 2028 is a start-up date rather than another revision. Even then, Russia remains the largest supplier. The gain is not independence. It is the ability to absorb a 90-day shock without rationing herders and haul trucks.

The adverse path is the one Dashpurev sketched. If Russian product stays tight for two or three years, and if Chinese diesel remains the swing barrel Beijing is least willing to spare, mining schedules slip, the import bill rises, and the political incentive is to trade deeper export dependence on China for emergency fuel. That bargain would be rational in a bad winter and costly for a decade. It would also make the third neighbor irrelevant to the one market in which Mongolia is already most exposed.

The October march did not reveal that landlocked autonomy is a fiction. Mongolia has spent thirty years building a foreign policy that keeps two neighbors from becoming a single fate, and the Indian refinery under construction and the Korean contract are evidence that the policy can be turned toward energy. What the march revealed is the unit in which that policy has to be scored. Eleven days of cover, a 97 percent supplier share, a refinery that at best covers a third of 2028 demand, and a southern neighbor that already buys the exports: those are the constraints. Third-neighbor sovereignty will be credible when a disruption in a Russian refinery changes a contract schedule rather than the length of a queue. Until then it is a strategy with the right partners and the wrong buffer.

BY Rahul Guhathakurta

Published Date:2026-10-03