Celltrion, Inc. (KRX:068270) reported ₩1,168.5bn of operating income for FY2025 against ₩492.0bn in FY2024. That is a 138% increase on 17% revenue growth, the kind of ratio that gets described as an inflection. Look at the cost line and a duller explanation appears first.
Cost of revenue in FY2025 was ₩1,695.5bn. In FY2024 it was ₩1,875.6bn. Costs went down 9.6% while revenue went up 17.0%. No amount of factory utilization, product mix or pricing discipline produces that combination in a single year. Something was in FY2024's cost of goods that stopped being there.
Celltrion absorbed Celltrion Healthcare, its former marketing affiliate, at the end of 2023. Before that, the manufacturer sold product to the affiliate and the affiliate sold it to hospitals and wholesalers. The structure had been criticized for years, because the transfer price between the two determined how profit split between two separately listed companies with overlapping owners.
When a merger like that closes, the acquirer restates the acquired company's assets to fair value. Inventory gets stepped up. As that inventory sells through, it flows into cost of goods at the stepped-up figure rather than at what it cost to make, and reported gross margin sags for however long the stock takes to clear. Then it stops. I should be clear that I'm inferring this from the shape of the numbers, not quoting a note that spells it out. But the shape is unambiguous: gross margin ran 48.3% in FY2023 and 47.3% in FY2024, then 59.3% in FY2025 and 62.0% in the second quarter of 2026. Quarterly, it was 51.1% in Q4 2024 and 64.2% by Q4 2025.
A merger drag rolling off is a good thing. It is not the same thing as the business getting more profitable, and the difference matters for anyone building forward numbers from FY2025 growth rates. There is nothing left to unwind. FY2026 has to earn its margin gain the ordinary way.
Two things, and both are real.
The first is operating leverage on the sales organization. Selling, general and administrative expense was ₩400.4bn in FY2023, then ₩1,189.7bn in FY2024 when the commercial arm arrived on the consolidated income statement, then ₩1,298.6bn in FY2025. As a share of revenue that is 33.4% in FY2024 falling to 31.2% in FY2025 and 29.6% in the second quarter of 2026. A biosimilar company with its own distribution has a mostly fixed selling cost against a growing product count. Adding Steqeyma or Yuflyma to a rep's bag costs far less than the first product did. That leverage should continue and it is the honest bull argument for margins from here.
The second is cash. Operating cash flow was ₩646.1bn for all of FY2025 against ₩1,168.5bn of operating income, a poor ratio, and the company burned cash overall after ₩860.4bn of investing outflows. The first half of 2026 looks different: ₩651.4bn on a cumulative basis through June, against ₩301.6bn in the first half of 2025. That is more than a doubling, and it means six months of 2026 produced as much operating cash as the whole of the prior year.
A note on reading those figures. Korean interim cash flow statements are cumulative from the start of the year, not discrete quarters, so the June number is the half and not the second quarter alone. Accounting arguments about inventory step-ups do not touch operating cash flow the way they touch gross margin. A doubling there is the strongest single fact in the recent numbers.
Inventory is the thing that could unwind it. Inventories stood at ₩2,940.5bn at the end of the second quarter against ₩2,712.1bn a year earlier and ₩2,803.5bn at year-end. That is more than two full quarters of cost of revenue sitting in a warehouse. Some of it is deliberate — biosimilar launches run to tender calendars and you cannot be short. Some of it is the residue of the merger. Either way, the working capital release that would make cash flow look permanently better has not happened yet.
FY2025 pretax income was ₩1,153.7bn against ₩576.1bn, a doubling. Net income went from ₩418.9bn to ₩1,031.5bn, up 146%. The gap is the tax rate: 10.6% in FY2025 against 27.3% in FY2024.
Run the counterfactual. Tax FY2025's pretax income at FY2024's rate and net income lands near ₩839bn instead of ₩1,031bn. So roughly ₩193bn of the ₩613bn net income increase, close to a third, came from the tax line rather than the business.
The quarterly detail is odder still. Income tax expense was negative ₩64.7bn in Q4 2025 and negative ₩10.9bn in Q2 2025. A negative tax charge usually means a deferred tax asset was recognized or a prior provision released. Neither is fraudulent, and Korean R&D and facility investment credits are generous. But a company that books a tax benefit in two of eight quarters does not have a settled effective rate, and 10.6% is not a rate to model forward. The first half of 2026 ran closer to 19%, on ₩168.8bn of tax against ₩889.9bn of pretax income.
One more thing that gets skipped. Total comprehensive income was ₩778.8bn in FY2025, below net income of ₩1,031.5bn. In Q2 2025 comprehensive income was negative ₩284.0bn against positive net income of ₩63.3bn. Celltrion holds most of its assets outside Korea in some form and translates them back, so the won's moves swing equity by hundreds of billions without touching the P&L.
That is why total equity fell from ₩17,580.1bn at the end of FY2024 to ₩17,352.5bn at the end of FY2025 despite ₩1,031.5bn of profit. Reported earnings and the change in book value have not agreed for two years. Investors who anchor on either alone will get a different company.
Take the other side properly. The pre-merger Celltrion earned gross margins in the high 50s and above, and the current 60-62% is not an aberration but a return to where a manufacturer with its own distribution should sit once the accounting noise clears. On that reading FY2024 was the anomaly, FY2025 was the normalization, and I'm treating a recovery to trend as if it were a trick.
The revenue growth is also not in dispute and it is accelerating. First-half 2026 revenue of ₩2,538.7bn is up about 41% on the prior year, and management has pointed to newer, higher-margin products — Zymfentra, Yuflyma, Steqeyma — passing 60% of the total. If mix is genuinely shifting toward the products with better economics, margin can hold at these levels for reasons that have nothing to do with a step-up ending.
And the second-half seasonality is real in this industry. European national tenders and year-end stocking concentrate in the back half, which is why a first half at 41% growth is not the ceiling for the year.
The FY2026 effective tax rate, first. If the full year lands near the first half's 19% rather than FY2025's 10.6%, then a meaningful piece of last year's earnings growth was borrowed and the year-on-year comparison in early 2027 will look worse than the business does.
Second, gross margin in the second half. The seasonal peak quarters should be the best margin quarters if mix is doing the work. Q4 2025 hit 64.2%. If Q4 2026 comes in at or above that on much higher revenue, the leverage argument is proved and the accounting argument is spent. If margin flattens around 60% while revenue climbs, the step-up explanation was carrying more of FY2025 than management's framing allowed.
Third, inventory days. Two quarters of cost of revenue in stock is defensible before a launch wave and hard to defend after one.
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.