POSCO Holdings Inc. (KRX:005490, NYSE:PKX) reported net income of ₩1,304.7bn for the first half of 2026, against ₩428.1bn a year earlier. Triple. It is also more than two and a half times what the group earned across the whole of 2025, which was ₩504.4bn.
That's a real number and the recovery underneath it is real too. But "profit tripled" is doing a lot of work in the coverage, and the half-year income statement is specific enough to say where the money actually came from. Operations supplied a bit over a third of it. The rest came from lines that don't repeat on command.
Pre-tax profit went from ₩737.6bn to ₩1,667.7bn, an improvement of ₩930.1bn. Four lines produced it.
Operating profit rose ₩350.2bn, from ₩1,175.6bn to ₩1,525.8bn. That's 38% of the total.
Other non-operating income and expense swung from negative ₩120.2bn to positive ₩287.8bn, a movement of ₩408.0bn, or 44%. Almost all of it is on the income side: other non-operating income was ₩691.5bn against ₩183.5bn, up ₩508.0bn, while the corresponding expense line rose only ₩99.9bn. The summary statements don't itemise what's in that ₩691.5bn. The notes to the half-year report would; I haven't read them, and I'm not going to guess at the split.
Net finance costs improved by ₩160.8bn, from negative ₩470.5bn to negative ₩309.8bn — 17% of the total. Worth noting how gross those numbers are: finance costs of ₩2,633.4bn and finance income of ₩2,323.6bn in a single half, at a group whose entire gross borrowings are ₩30.4tn. Interest on that debt wouldn't be a quarter of either figure. These lines are dominated by currency and derivative movements passing through in both directions, and the ₩160.8bn is a residual between two much larger numbers.
Equity-method income contributed ₩11.0bn. Rounding.
Then tax. The effective rate fell from 42.0% to 21.8%. Tax expense was ₩363.0bn on ₩1,667.7bn of pre-tax profit. Had last year's rate applied, the charge would have been roughly ₩700.4bn and net income about ₩967.3bn instead of ₩1,304.7bn. The half-year summary doesn't explain the drop, and a 42% effective rate in the prior period was itself unusually high — Korean statutory rates are well below that, and a rate that far above statutory usually means non-deductible losses somewhere in a consolidated group. So part of the decline is probably last year's number normalising rather than this year's being generous. Either way, a 20-point move in the effective rate is not an operating result.
Korean coverage of the quarter framed the environment as a weak won and Middle Eastern energy supply risk, and the accounts back the currency half of that up.
In other comprehensive income, foreign operations translation gains were ₩639.2bn for the half against a loss of ₩566.8bn a year earlier — a ₩1.21tn swing that never touches reported profit but tells you which way the won moved. POSCO Holdings owns lithium in Argentina, nickel and coal interests in Australia, and a trading arm with large overseas balances. A weaker won revalues all of it upward.
The same move is a cost on the other side. POSCO imports essentially all of the iron ore and coking coal it consumes, priced in dollars, and a weaker won raises the won cost of every tonne. Both effects are inside these accounts. The one that shows up in the non-operating lines helped; the one that shows up in cost of sales hurt.
Which makes the gross margin the more interesting number. Cost of sales rose 5.2% while revenue rose 6.1%, so gross profit grew 16.7% to ₩3,175.8bn and the gross margin went from 7.78% to 8.55%. Seventy-seven basis points, achieved while the currency was working against the input side of the ledger. That's a genuine operational result, and it's the part of this half I'd take most seriously.
Three things in this half are unambiguously operational, and none of them is "profit tripled."
Revenue grew for the first time in years. It was ₩77.13tn in 2023, ₩72.69tn in 2024 and ₩69.09tn in 2025 — two consecutive annual declines totalling 10.4%. The first half of 2026 came in at ₩37.13tn against ₩34.99tn, up 6.1%, and the second quarter at ₩19.26tn was 7.7% ahead of the first. The top line turned.
The subsidiaries turned with it. Profit attributable to non-controlling interests went from negative ₩34.1bn to positive ₩152.7bn. Since POSCO's minorities sit mostly in the separately listed POSCO Future M and POSCO International, that swing is direct evidence that the battery materials and trading businesses moved from loss to profit rather than the parent's steel operation carrying everything. The company has said as much: battery materials returned to profit, the lithium operation turned profitable, and POSCO International posted a record quarterly operating profit.
And the new electric arc furnace at Gwangyang started running during the half. That's capacity that changes the cost structure rather than just adding tonnes.
Quarterly, operating profit was ₩706.8bn in the first quarter and ₩819.0bn in the second, a 15.9% sequential gain, with the margin moving from 3.95% to 4.25%. Two consecutive quarters of improvement in the line that matters.
The strongest objection to my framing is that decomposing a profit into "operating" and "everything else" is arbitrary for a group like this. POSCO Holdings is a holding company whose businesses include resource extraction and commodity trading. Currency movements and asset revaluations aren't noise sitting on top of the business — for POSCO International in particular they substantially are the business. Insisting that only the operating line counts imposes a manufacturer's income statement on something that isn't purely a manufacturer.
That's fair, and it's why I'd point at gross margin rather than at operating profit as the cleanest read.
The risk running the other way is the one the headline invites. A reader who takes ₩1,304.7bn, annualises it, and puts it against a ₩25.99tn market capitalisation gets a single-digit multiple and concludes the stock is obviously cheap. First-half basic earnings per share were ₩15,233; double it and the multiple is about 10.8x. But roughly ₩580bn of that first-half profit came from non-operating income and a tax rate that halved. Strip those toward normal and the run-rate multiple is closer to the mid-teens — which, notably, is about where the enterprise value multiple already sits.
Then there is the thing the income statement doesn't capture at all. Operating cash flow in the half was ₩708.6bn, against ₩1,304.7bn of reported profit and ₩2,118.7bn of capital spending. The profit tripled; the cash did not follow.
The gross margin in the third-quarter report, against 8.55% for the half and the implied second-quarter improvement. That is the number that separates a cost-structure change from a currency effect, and it's the one that has to hold for any of this to compound.
Second, whether the effective tax rate stays near 21.8% or reverts. A nine-month rate back above 30% takes several hundred billion won straight out of the reported figure without anything changing in the business.
Third, the other non-operating income line against ₩691.5bn. If the third quarter adds another few hundred billion from the same undisclosed source, it's structural to how this group earns and I've mislabelled it. If it drops back toward the ₩183.5bn of a year ago, the second half will look like a sharp deceleration that isn't 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.