Half of every won that walked in Leeno Industrial's (KOSDAQ:058470) door last quarter walked out as operating profit. Revenue was ₩143.2bn, operating income ₩73.5bn, margin 51.3%. Software companies post numbers like that. Companies that machine tiny metal pins in Busan generally do not.
The interesting part is not the level. It is the direction, and how little the direction has cared about the semiconductor cycle.
Run the operating margin down the filed history and the shape is almost monotonic. 2015: 36.2%. 2019: 37.7%. 2020: 38.7%. 2021: 41.8%. 2022: 42.4%. 2023: 44.8%. 2024: 44.6%. 2025: 47.5%. First half of 2026: 49.7%, with the second quarter alone at 51.3%.
Look at 2023 again. Revenue dropped from ₩322.4bn to ₩255.6bn, a 21% fall, in the worst memory downturn in a decade. The margin went up. That is the single most useful fact about this company, and it separates Leeno from almost every other name in the Korean semiconductor supply chain. Equipment makers see margins collapse in a down year because they carry fixed cost against lumpy tool shipments. Leeno sells consumables. Chips still get tested in a bad year, just fewer of them, and the sockets still wear out.
Two things drive it. The first is mix: as chips get harder to test, sockets get more expensive per unit and the design work behind them gets more specialized. The second is that the cost base barely moves. Revenue grew about 3.7x between 2015 and 2025. SG&A over the same span went from ₩9.6bn to ₩17.4bn, about 1.8x. In the second quarter of 2026 the company spent ₩5.4bn on SG&A against ₩143.2bn of revenue, which is 3.7 cents on the dollar. Most industrials would take that number and frame it.
Gross margin tells the same story more quietly: 45.8% in 2015, 52.2% in 2025, 55.1% in the second quarter of 2026. So roughly nine points of the fifteen-point operating margin improvement came from the factory, and the rest from not adding overhead.
Leeno files no consolidated statements. It has no subsidiaries, so everything here is separate parent-only accounting from DART, which for once makes the analysis simpler rather than harder. What you see is the whole company.
And what you see is unusual. Total liabilities of ₩80.1bn against total assets of ₩856.9bn. Equity of ₩776.8bn, an equity ratio near 91%. Cash of ₩86.8bn, which by itself exceeds every liability on the sheet. Finance costs in 2025 came to ₩6.5m against finance income of ₩9.6bn, a ratio of roughly 1,500 to one. There is effectively no debt and there has not been for the whole filed history.
That matters for margins in a way people skip past. Pretax income in 2025 was ₩195.9bn against ₩177.0bn of operating income, because interest on idle cash adds rather than subtracts. A leveraged competitor running the same factory at the same gross margin would report meaningfully less at the bottom.
Inventory is lean as well: ₩20.5bn at the end of June against ₩64.3bn of quarterly cost of revenue. The 2025 full-year balance of ₩15.5bn worked out to about a month of cost of sales. Custom sockets get built against orders, not into a warehouse, and that keeps working capital small even as revenue compounds.
Leeno is spending roughly ₩200bn (~$145m at about 1,400 won to the dollar, approximate) on a new facility in Busan's Eco Delta City, targeted for the second half of 2026 and reported to more than double capacity. The company's own value-up disclosure on March 26 names it directly and ties it to funding shareholder returns.
The capex is already visible. Purchases of property, plant and equipment ran ₩11.6bn in all of 2024, then ₩60.0bn in 2025, then ₩41.4bn in the first half of 2026 alone. PP&E on the balance sheet went from ₩181.4bn at the end of 2024 to ₩262.7bn at the end of June. That is a company that spent almost nothing for years and has now added roughly 45% to its fixed asset base in eighteen months.
Here is the tension. Doubling capacity means doubling the depreciation that runs through cost of revenue. If demand fills the plant, the operating margin holds or goes higher because SG&A stays flat and the gross margin absorbs the depreciation on much larger volume. If demand arrives two years late, the same depreciation lands on today's revenue and the gross margin gives back several points before anything is wrong with the business.
That is a normal manufacturing risk. It is worth stating plainly because it is the first time in this company's recent history that the margin has faced a real test from its own decisions rather than from the market.
The bear case does not need the plant to fail. It only needs the margin to be a cycle artifact.
Test socket demand is tied to new chip designs and to test intensity per chip. Both are running hot right now: advanced packaging, higher pin counts, tighter pitches, and mobile processors moving to leading-edge nodes all raise the value of the socket per chip tested. That is a genuine structural tailwind. It is also the kind of tailwind that gets extrapolated at exactly the wrong moment. If design starts slow down for a few quarters, the consumable revenue slows with them, and a 51% margin on a smaller base is a smaller number.
Competition deserves more weight than it usually gets. Leeno's advantage is turnaround time and design capability on customized parts, not a patent thicket. Korean and Japanese rivals compete for the same sockets, and a 51% margin is a standing invitation to price against. Nothing in the filings suggests pricing pressure right now. Nothing in the filings would show it early either.
There is also a concentration question the public record does not answer well. A supplier of custom test parts to a small number of large chipmakers has customer concentration by construction. The half-year report does not break out customer share in a form that settles it, so I won't pretend to know how exposed the revenue is to any single account.
Gross margin is the tell, not operating margin. Once the new plant starts depreciating, watch the gross margin line specifically. Second-quarter 2026 gross margin was 55.1%. If it holds above roughly 52% through the first couple of quarters after the plant comes online, the capacity is being filled and the operating margin will look after itself. If it slips into the high forties while revenue keeps growing, the plant is running ahead of demand and the 51% quarter was the top.
The completion itself is the other marker. The value-up disclosure says second half of 2026, which leaves roughly four months. Whether that date holds, and what the third-quarter report in November says about capex still to come, will tell you more than any commentary about AI demand.
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.