Korea Electric Power Corporation (KRX:015760, NYSE:KEP) sold ₩46,317.3bn of electricity in the first half of 2026, essentially the same as the ₩46,174.1bn it sold a year earlier. Operating profit fell 16.6%, from ₩5,889.5bn to ₩4,912.7bn. Net income fell 21.0%, from ₩3,538.1bn to ₩2,796.5bn.
Flat revenue, profit down a fifth. For most companies that would prompt a question about costs, competition or management. For KEPCO it prompts a different one, and it's the only question that matters here: what happened to the fuel price, and what happened to the tariff.
Fuel went up. The tariff did not.
KEPCO buys coal, gas and uranium at world prices, in dollars. It sells electricity to every household and business in Korea at a price approved by the Korean government. It has no ability to set that price, no formula that adjusts it automatically to input costs, and no rate-of-return framework of the kind a US utility operates under. American utilities are regulated on the principle that they should be able to earn their cost of capital. KEPCO has no such guarantee and never has.
The consequence shows up as an income statement that swings wildly while the business does nothing unusual.
Revenue was ₩88,219.5bn in 2023, ₩93,398.9bn in 2024 and ₩97,429.3bn in 2025 — up 10.4% across two years, which is roughly what you'd expect from tariff increases and slowly growing demand.
Operating profit over the same three years was negative ₩4,541.6bn, then positive ₩8,364.7bn, then positive ₩13,490.6bn. A swing of ₩18.0tn on a revenue line that moved less than ₩10tn.
No operational decision produces that. What produced it was the gap between a fuel bill and a regulated price, first widening catastrophically after 2021 and then closing as tariffs were raised and fuel came off its peak.
The first half of 2026 is the same mechanism running backwards a little. Korean coverage attributes the second-quarter weakness to energy prices rising in the wake of Middle East disruption, with international thermal coal in particular pushing the fuel bill up, while power sales volumes fell only slightly. Operating expenses for the half were ₩41,404.6bn against ₩46,317.3bn of revenue.
Some Korean coverage of these results reported first-half revenue as ₩26.3tn. The half-year report filed with the Financial Supervisory Service says ₩46,317,339m — ₩46.32tn.
The filing is right and the arithmetic proves it: the same articles report operating expenses of ₩41,404.6bn and operating profit of ₩4,912.7bn, and those two only reconcile with ₩46.32tn of revenue. It looks like a transposition somewhere in the chain. Worth flagging because a US reader working from translated coverage would have no way to catch it.
Three details from the accounts make the structure concrete.
Inventories were ₩11,139.7bn at the end of June, up from ₩10,152.7bn. That's coal and fuel oil sitting in stockpiles, and it rose 9.7% in a half when revenue was flat. A company that could pass costs through wouldn't much care what its fuel inventory was worth. KEPCO's entire profit and loss is decided by the spread between what that pile cost and what the government lets it charge for the electricity made from it.
Non-current provisions were ₩29,149.8bn. That's principally nuclear decommissioning and spent fuel — a liability accruing on a schedule measured in decades, sitting on the balance sheet of a company whose revenue is set annually by political decision.
And equity accounted for using the equity method ran to ₩7,192.1bn in associates plus ₩7,018.4bn in joint ventures. KEPCO has significant overseas power projects, which are the one part of the business priced by markets rather than by Seoul, and they are a small fraction of a ₩262.6tn balance sheet.
At ₩34,800 the market capitalisation is ₩22,340.3bn. Equity attributable to the parent was ₩50,675.8bn at the end of June, so the shares trade at 0.44x book. Annualise first-half net income of ₩2,796.5bn and the multiple is about 4.0x; measure against 2025's ₩8,666.7bn and it's about 2.6x.
Those are numbers you normally see attached to a company the market believes is about to lose money. The market may be right, and it wouldn't require a mistake to be right — KEPCO lost ₩4.7tn in 2023 for exactly the reasons that could recur.
The shares are at ₩34,800 against a 52-week range of ₩30,850 to ₩67,900. They have roughly halved from the high.
The bull case is that the cycle has turned in KEPCO's favour and the market hasn't repriced it. Tariffs were raised substantially between 2022 and 2024 and have not been reversed. The nuclear fleet's utilisation has been running high, which lowers the average fuel cost per unit generated. A company earning ₩13.5tn of operating profit in 2025 and trading at ₩22.3tn of market capitalisation is priced as though 2025 was a fluke.
The bear case does not need the fuel price to spike. It only needs the tariff to stay where it is while costs drift up, which is precisely what the first half of 2026 showed. And there is a specific political asymmetry: raising electricity prices in Korea is unpopular and gets deferred, while lowering them when fuel is cheap is popular and happens quickly. Over a full cycle, that asymmetry is a transfer from KEPCO's shareholders to Korean consumers, and no amount of operational competence changes it.
There's a third reading I'd hold alongside both. KEPCO's earnings are so completely determined by two exogenous prices that the usual analytical apparatus — margins, returns on capital, management quality — has almost no predictive value. Anyone building a model of this company is really building a model of Korean energy policy, and would be better off doing that explicitly.
The 2026 tariff decision, which is expected at the end of December. That is the single most important scheduled event for this company, it will be made by the government rather than the board, and it will determine the 2027 income statement more than anything KEPCO does between now and then.
Second, third-quarter fuel costs. The third quarter is Korea's peak demand quarter and the one where a fuel-price move does the most damage or the most good. Operating expenses of ₩41,404.6bn for the half is the base to measure against.
Third, the inventory line against ₩11,139.7bn. A continued build while volumes are flat tells you the fuel bill is still rising, which is the thing the December decision will or won't compensate for.
kstock reads DART every morning and writes up what moved — the contract, the buyback, the number that does not add up. The daily post and a Saturday roundup, by email.