POSCO Holdings (KRX:005490) sold ₩2.50tn of two listed subsidiaries on August 7. The money settles Monday, September 7. Read the two disposal filings down to the last box and the deal changes shape. POSCO signed a three-year price return swap on every share it sold. If those shares fetch less than the sale price when the swap unwinds, POSCO pays the difference. If they fetch more, POSCO collects it. It also stopped selling at a bare majority of each company, down to the last share. It still controls both.
The cash is real. The exit is not.
One note on currency first. POSCO reported its Louisiana investment as both $582m and ₩858.6bn. That pair implies about 1,475 won to the dollar, and I use that rate where I convert. Treat it as approximate.
The bigger trade is POSCO International, the group's trading and energy arm. POSCO Holdings sold 36,434,963 shares at ₩55,400 each. That was the prior day's close. The sale raised ₩2,018,496,950,200. That leaves it holding 87,961,395 shares. POSCO International has 175,922,788 shares outstanding. Half of that is 87,961,394.
One share.
The smaller trade is POSCO DX, the group's information technology and engineering unit. POSCO Holdings sold 23,385,917 shares at ₩20,600 for ₩481,749,890,200. It keeps 76,017,365 shares out of 152,034,729. That is the smallest whole number above half.
Both filings report the outcome as "50%." Both are really a bare majority, sized to the share. POSCO International went from 70.7% down to that number. POSCO DX went from 65.4%. Whoever built these trades counted.
The reason to count is consolidation. A majority of the shares outstanding is the cheapest way to keep a subsidiary's revenue and profit inside the group accounts. POSCO International is not a small thing to lose. The summary table inside the filing puts its 2025 revenue at ₩32.37tn and its 2025 net income at ₩636.8bn. POSCO Holdings reported ₩504bn of consolidated net income that year. The trading subsidiary earned more than the entire group did. The group eliminates intra-group sales on consolidation. So the revenue would not simply disappear from the top line. The profit is a different matter.
Both filings put the swap in the final box, under other matters relevant to an investment decision. The terms match each other. Three years. The underlying is the exact block of shares sold. The reference price is the price those shares were sold at, ₩55,400 for POSCO International and ₩20,600 for POSCO DX. At settlement, the two sides calculate what the shares actually fetch, net of costs. They settle the difference between that amount and the reference price.
So POSCO sold ₩2.50tn of stock and kept ₩2.50tn of price exposure to it. What changed is the form. The shares left as equity. The risk stayed as a derivative.
Earlier in the same form, both filings answer the question about a put option or similar contract with "no." That answer is technically correct. The distinction also matters. A put would cap POSCO's loss and cost a premium up front. This swap caps nothing. POSCO keeps the whole outcome, upside and downside, for three years.
Neither filing says what POSCO pays the counterparty to carry ₩2.50tn of stock for that long. Korean press has named NH Investment & Securities as the swap counterparty. The filings name no buyer and price nothing. That missing number is the one an investor would most want to see.
The exposure is not small next to POSCO's own equity value. The company's market capitalisation was ₩26.7 trillion at Friday's close. So ₩2.50tn is roughly 9% of it. A 20% fall in those two subsidiaries over three years would cost about ₩500bn at settlement. POSCO's entire 2025 net income was ₩504bn.
Ask why a company sells ₩2.50tn of anything.
Across the five years from 2021 through 2025, POSCO generated ₩29.85tn of operating cash flow. It spent ₩28.06tn on property, plant and equipment. That leaves ₩1.79tn. Dividends over the same five years took ₩5.10tn. Debt covered the difference. Total liabilities went from ₩36.67tn at the end of 2021 to ₩42.81tn at the end of 2025.
The spending bought less each year rather than more. Operating income was ₩9.24tn in 2021 and ₩1.83tn in 2025. Operating margin fell from 12.1% to 2.6%. Revenue fell too, from ₩76.33tn to ₩69.09tn. In 2025 POSCO earned ₩504bn on ₩62.38tn of total equity, a return of 0.8%. Retained earnings stood at ₩53.65tn at the end of June. That is below the ₩53.86tn of December 2023. Two and a half years of trading added nothing to the pile.
The first half of 2026 did not close the gap. Operating cash flow was ₩709bn. Capex was ₩2.12tn and dividends took ₩405bn. Total liabilities rose ₩2.73tn in six months while cash fell ₩703bn.
Now run the proceeds against that. At the first half's rate, ₩2.50tn covers about eight months of the gap between what operations produce and what plant and dividends consume. That is a lot of money and a short amount of time.
On September 4 POSCO's partners broke ground on a $5.8bn electric-arc furnace mill in Ascension Parish, Louisiana. Hyundai Steel leads it with 50%. Hyundai Motor and Kia hold 15% each. POSCO holds 20%, reported at $582m, or ₩858.6bn. The mill is designed for 2.7 million tonnes a year of sheet for American and Mexican car plants. Commercial operation is set for 2029. Those figures come from the partners' announcements and press reports. I have not seen them in a POSCO Holdings filing.
Set the two numbers side by side. POSCO's entire American steel commitment is ₩858.6bn. It just raised ₩2.50tn. The sale is close to three times the project.
Neither disposal filing names a use for the money. Both give the purpose in the same words: resolving the holding-company discount and securing funds for strategic investment. That is a category, not a project. POSCO also published a corporate value-up plan in July. In Korea that is a voluntary filing in which a company sets out its own targets for returns and shareholder payouts. Reports of that plan put the asset restructuring target at ₩3.5tn of cash by 2028. They report ₩475bn booked in the first half of 2026 across twelve projects. One August afternoon raised more than five times that half-year figure.
They may simply have sold well. Second-quarter operating income was ₩819bn. That was the highest of the eight quarters back to the third quarter of 2024. Second-quarter revenue of ₩19.26tn was also the highest of the eight. First-half operating income of ₩1.53tn is up 30% from ₩1.18tn a year earlier. If steel keeps recovering, both subsidiaries should do fine over three years. The swap then settles in POSCO's favour, and POSCO will have raised ₩2.50tn for little more than a financing fee.
The structural argument is fair too, and management makes it right in the filing. A holding company whose value sits mostly in listed subsidiaries trades below the sum of its parts, because investors will not pay twice for the same asset. Cutting two stakes and moving the cash into operating assets is a standard answer to that. It is why both filings lead with the holding-company discount.
The swap is also disclosed in plain language, on the day the board approved it. Nothing here is buried.
The risk I see is narrower than "too much leverage." POSCO has traded an equity stake for cash plus a three-year obligation whose size depends on two share prices it does not control. Against 2025 net income of ₩504bn, a ₩500bn settlement is not a rounding error. And the recovery has to hold for three years, not two quarters. Steel does not offer three-year visibility.
The third-quarter report is the document that answers this. Four lines to read.
Where the ₩2.50tn lands in the cash flow statement. Investing activities means the accountants treated it as a disposal. Financing activities, or shares still sitting on the balance sheet, means they read the swap the way I do, as borrowing against stock rather than a sale.
Whether POSCO International and POSCO DX are still consolidated, and whether non-controlling interests jump. A bare majority should hold that consolidation. The word "should" is doing some work there.
The derivative note. Look for the swap's fair value and, more usefully, its cost. Neither August filing disclosed a fee.
Capex. POSCO spent ₩2.12tn on plant in the first half against ₩709bn of operating cash. If the proceeds fund that same run rate into 2027, the company will be back for more. If capex comes down instead, the sale bought time to fix the return.
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.