Between the end of September and the end of December 2025, Samsung Biologics Co., Ltd. (KRX:207940) shrank by ₩7,276.6bn. Total assets went from ₩18,337.3bn to ₩11,060.7bn. Intangible assets went from ₩5,391.8bn to ₩60.2bn. Total equity went from ₩12,179.4bn to ₩7,451.1bn. Nothing was written off. Samsung Bioepis, the biosimilar developer, left the group into a separate holding company.
What remains is a pure contract manufacturer, which is what the company said it wanted. It is also a company whose reported history no longer describes it, and the filings make that harder to see than it should be.
Worth stating plainly, because the accounting consequences obscure a genuinely sound commercial decision. Samsung Biologics manufactures drugs for large pharmaceutical companies under contract. Samsung Bioepis develops biosimilars — near-copies of biologic drugs — and those biosimilars compete directly with the branded products some of those same customers sell.
A pharmaceutical company deciding where to place a manufacturing contract has to weigh whether it wants its process know-how sitting inside a group whose other arm is building a competing version of its molecule. Samsung Biologics has said the separation addresses exactly that conflict. Lonza and Boehringer Ingelheim, its two largest rivals, do not have this problem in the same form. Removing it is worth something that will never appear as a line item.
Share capital fell from ₩177,935,000,000 to ₩115,727,377,500 in the same quarter. At the ₩2,500 par value stated in the company's recent rights offering filing, that is a share count going from 71,174,000 to 46,290,951 — a 35% reduction.
That is how a Korean horizontal spin-off (인적분할) works when the parent consolidates its remaining stock. Shareholders received shares in the new holding company and their Samsung Biologics holding was reduced. Nobody lost economic value at the moment of the split. But it means every earnings-per-share, book-value-per-share and dividend-per-share figure published for Samsung Biologics before the fourth quarter of 2025 refers to a company with 54% more shares outstanding and a different asset base. Charting that series produces a discontinuity that looks like an event and isn't.
Anyone valuing the shares off a five-year EPS trend is comparing two different securities.
Here is the specific trap. Take the FY2025 income statement as filed and subtract the nine-month cumulative figures, which is how quarterly numbers get built for Korean filers that report cumulatively. For the fourth quarter of 2025 you get revenue of ₩308,572,500,609 and selling, general and administrative expense of negative ₩292,106,322,004.
Neither is a real quarter. Revenue of ₩308.6bn would be a quarter of the first-quarter figure, from a company reporting full utilization across Plants 1 through 4. Overhead cannot be negative. What has happened is that the annual figure and the interim figures are not on the same basis — one reflects the deconsolidation and one does not — and subtracting them produces a residual with no meaning.
The knock-on is that year-on-year quarterly growth also breaks. Second-quarter 2026 revenue of ₩1,320.9bn against the ₩1,289.9bn shown for the second quarter of 2025 works out to 2.4% growth. The company reports 30% growth for the same quarter. Both are defensible. The 2.4% compares a CDMO-only quarter against a quarter that included Bioepis; the 30% compares like with like. The filed series gives you the first and only the disclosure gives you the second.
For a US reader used to Regulation S-X requiring restated comparatives for discontinued operations, this is the thing to watch for in Korean filings. Comparatives are not always restated in the interim series even when the annual is.
Now the part that matters for the business. Second-quarter 2026 operating income was ₩586.4bn on ₩1,320.9bn of revenue, a 44.4% operating margin, against ₩475.6bn on ₩1,289.9bn a year earlier, or 36.9%. A 7.5-point improvement.
Split it. Gross margin was 53.0% in the second quarter of 2026 and 55.7% a year earlier — down, not up. Selling, general and administrative expense was ₩113.4bn, 8.6% of revenue, against ₩243.4bn, or 18.9%. The first quarter of 2026 ran even leaner at 7.9%.
All of the operating margin gain and then some came out of overhead, and the overhead left with Bioepis. A biosimilar developer carries commercial infrastructure, marketing spend and clinical costs. A contract manufacturer does not sell to patients; it sells to a few dozen procurement departments. Eight to nine percent of revenue in overhead is what that business should look like.
The gross margin drift is worth watching. It moved from 55.7% to 54.1% to 53.0% across three quarters. A 2.7-point decline over a year is not alarming and could be mix, currency, or the ramp of Plant 5, which entered full operation in April 2025 and would have carried startup costs. It is also the only margin line the company now controls through its own operations, so it deserves more attention than it usually gets.
The clearest way to see what changed: intangible assets are ₩76.8bn of ₩12,652.6bn in total assets, six-tenths of one percent. Property, plant and equipment is ₩6,468.0bn, or 51%. Before the split, at the end of FY2024, intangibles were ₩5,565.3bn of ₩17,336.3bn, or 32%.
Samsung Biologics is now a company you can walk around. Its book value is bioreactors, buildings and inventory, not acquired product rights. For a business whose competitive position rests on installed capacity — 845,000 litres across Songdo and Rockville by the company's account — that is the honest balance sheet.
It also means impairment risk has largely left. Whatever else goes wrong here, it will not be a goodwill writedown.
The counter-argument is that Samsung Biologics gave away the part of the business with the highest long-term return. Bioepis owns products. A CDMO rents capacity. Product economics scale without proportional capital; manufacturing economics do not. Every additional won of Samsung Biologics revenue from here requires roughly the capital intensity of the last one, which is why purchases of property, plant and equipment ran ₩1,391.7bn in FY2025 and ₩1,303.6bn in FY2024.
A 44% operating margin on a business that must keep spending 30% of revenue on plant is a different animal from a 44% margin on a business that doesn't. The free cash conversion tells that story: FY2025 operating cash flow of ₩2,247.8bn against ₩1,862.4bn of investing outflows left very little.
The other objection is timing. The company separated from biosimilars in the same year the FDA began dismantling the comparative efficacy requirement, which lowered the cost of building biosimilars considerably. Whether Samsung sold the wrong half at the wrong moment is a question the two share prices will settle over years, not quarters.
The FY2026 annual report will give the first clean full year of the CDMO alone, with comparatives that should finally be restated on one basis. Until then, build models from the four 2026 quarters and discard everything before.
Two specific numbers when it arrives. Gross margin for the full year, against the 53-56% band the recent quarters have traced — that is the pure manufacturing spread with no accounting boundary running through it. And capital expenditure, which will show whether Plant 6 has been committed. The company has said it is reviewing construction at its second Bio Campus while stabilizing Rockville. A capex number materially above FY2025's ₩1,391.7bn means the review ended in a yes.
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.