In FY2023, Hanmi Semiconductor Co., Ltd. (KRX:042700) generated ₩34,571,410,992 of operating income. It reported net income of ₩267,167,515,858.
Net income was more than seven times operating income. Pretax income was ₩345.3bn, so roughly ₩311bn arrived from somewhere below the operating line.
In FY2024 the company generated ₩255,391,604,854 of operating income — seven times more than the year before — and reported net income of ₩152,614,498,001. Less than in FY2023.
Two consecutive years in which the direction of net income had nothing to do with the direction of operating profit.
Take 2024 quarter by quarter.
First quarter: operating income ₩28.7bn, pretax income ₩90.5bn. A non-operating gain of about ₩62bn.
Second quarter: operating income ₩55.4bn, pretax income negative ₩15.3bn. A non-operating loss of about ₩71bn, which turned a profitable quarter into a net loss of ₩11.8bn.
Third quarter: operating income ₩99.3bn, pretax income ₩51.3bn. Another ₩48bn of non-operating drag.
Fourth quarter: operating income ₩71.9bn, pretax ₩71.8bn. Roughly neutral.
So in a single year, non-operating items added ₩62bn, subtracted ₩71bn, subtracted ₩48bn and did nothing. Anyone building a quarterly model of this company off net income was modelling noise.
The reported finance income and finance costs do not explain it either. Finance income was ₩34.5bn and finance costs ₩6.1bn across FY2024 — nowhere near the swings involved. So the movement sits in other non-operating income and expense, which the summary statements do not itemise.
I am not going to guess a specific transaction, but the shape narrows the possibilities.
A ₩311bn non-operating gain in FY2023, at a company whose entire operating income that year was ₩34.6bn and whose total equity was ₩571.9bn, is a very large one-off. Non-current assets jumped from ₩185.9bn at the end of FY2022 to ₩404.4bn at the end of FY2023, and investing activities produced a ₩84.6bn inflow. That combination is consistent with a disposal or a revaluation of a holding.
The 2024 swings — gains and losses of ₩50bn to ₩70bn alternating quarter to quarter — look more like fair value movements on financial assets held at fair value through profit or loss. Assets in that category get marked every quarter with the change running through the income statement, which produces exactly this pattern.
Both explanations point to the same conclusion for an investor: the non-operating line reflects what Hanmi owns rather than what Hanmi does, and the notes to the financial statements are the only place to find out what it owns.
Last year the relationship behaved. Operating income of ₩251.4bn produced pretax income of ₩278.4bn and net income of ₩214.0bn. Quarterly, the gaps were small: first quarter operating ₩69.6bn against pretax ₩70.4bn, second ₩86.3bn against ₩84.6bn, third ₩67.8bn against ₩88.3bn, fourth ₩27.6bn against ₩35.1bn.
Whatever was moving in 2023 and 2024 stopped moving, or was sold. That makes FY2025 a cleaner base for forecasting than either prior year, and it means the historical earnings series should be treated as three different things rather than a trend.
It also means the growth rates people quote for this company need care. Net income went from ₩267.2bn in FY2023 to ₩152.6bn in FY2024 to ₩214.0bn in FY2025 — down then up. Operating income went from ₩34.6bn to ₩255.4bn to ₩251.4bn — up sharply then flat. Those describe completely different companies and only the second one is the equipment business.
The current use of capital is more conventional and easier to read.
Property, plant and equipment rose from ₩126.5bn at the end of FY2023 to ₩238.1bn at the end of FY2025, with purchases of ₩27.5bn, ₩53.5bn and ₩74.6bn across the three years. The company filed decisions to acquire tangible fixed assets in June and again on 18 August 2026, and a decision to acquire a stake in another company on 12 June.
Dividends have risen alongside: ₩19.5bn paid in FY2023, ₩40.5bn in FY2024 and ₩68.3bn in FY2025 — about 32% of last year's net income. FY2024 financing outflows of ₩231.7bn against ₩40.5bn of dividends imply a substantial share repurchase in that year as well.
So Hanmi is building capacity, buying something, and returning cash, from a balance sheet with ₩276.2bn of cash and only ₩123.0bn of total liabilities.
The reasonable objection is that non-operating volatility is a presentation problem, not a business problem. Hanmi's TC bonders are selling; second-quarter 2026 revenue reached ₩251.1bn with a 51.9% operating margin, both records. If the operating line is clean and growing, an investor can simply use it and ignore the rest.
That is largely right, and it is what I would do. The caveat is that non-operating items are still real money. A ₩311bn gain in FY2023 was ₩311bn, and a ₩71bn loss in one quarter of 2024 was a real reduction in shareholders' equity. Ignoring the line for forecasting is correct. Ignoring it for valuation is not, because whatever produces those swings is an asset the company owns and you are buying.
The other non-operating income and expense note in the FY2025 annual report, which itemises what moved in each period. Three years of that table would explain the whole pattern in about five minutes of reading.
Second, whether the volatility stays gone. FY2025 was clean. If 2026 quarters show pretax income tracking operating income closely, the issue is historical and the earnings series can be used from FY2025 forward.
Third, what the 12 June stake acquisition was and how it will be accounted for. If it goes into a fair-value-through-profit-and-loss category, the volatility comes back — and this time an investor will at least know its name.
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.