SK hynix just reported the best quarter in its history — $42.8bn of operating profit on $56.0bn of revenue — and the stock has lost 51% from its June peak of ₩2,919,000 on 22 June. The explanations in circulation do not survive the filings. What does survive is a line item below operating profit, where the company booked more income from outside its business than inside it.
🔑 Key Takeaways
- Q2 non-operating income was $43.9bn — larger than the $42.8bn operating profit. Pre-tax income (₩122.7tn) came in at more than double operating profit (₩60.5tn). The same line was ₩3.3tn for all of 2025. This is subtraction between two reported figures rather than an estimate — though both are the company’s unaudited provisional numbers, pending the half-year report.
- The size matches a mark-up on the KIOXIA stake — but that attribution is our inference, and the half-year notes are the document that settles it. If it is right, part of it reverses in Q3; the Q1 cash flow shows some of the position was already being converted to cash.
- The cheap multiple is half non-operating. Annualised first-half earnings give 3.9×. On operating profit alone, taxed at Korea’s 26.4% statutory rate, it is 7.2×. Trailing 2025 is 22.9×. All three are honest; they answer different questions.
- SK hynix carries NAND two ways. Solidigm is a consolidated subsidiary and the KIOXIA stake is a shareholding. The company does not disclose Solidigm’s contribution separately, so how the 22-point gap to Micron splits between them is not knowable from filings.
- The July Nasdaq listing issued new shares. 17,790,000 of them, $26.5bn, struck at ₩2,242,301 — 58% above the 7 August 2026 close. Dilution 2.50%.
- The capacity to pay more was built in March and has not been used. The AGM converted ₩4.08tn of capital surplus into distributable profit expressly to fund returns. Since then the company has distributed 13.2% of it, at a quarterly dividend of ₩375 that has not changed since 2025. The 7 August filing commits to announcing additional measures during Q3 — and across 767 filings since January 2025 the company has never filed a Value-Up plan, so whatever arrives would be its first.
SK hynix (000660.KS; ADRs SKHY) is the world’s largest supplier of high-bandwidth memory, the DRAM that sits beside AI accelerators. In the second quarter of 2026 it turned that position into what appears to be the largest quarterly operating profit any Korean company has reported — Samsung Electronics’ entire FY2025 operating profit was ₩43.6tn against SK hynix’s ₩60.5tn in this one quarter. The share price responded by halving.
Two explanations circulate in English-language coverage: forced selling by leveraged holders, and a governance discount born of thin shareholder returns. The first is not testable from filings — Korean disclosure does not identify leveraged holders — so we set it aside rather than pretend to settle it. The second is testable, and what the filings show is more interesting than either yes or no. What the filings do contain is a line item that headline coverage skips.
🏢 What is SK hynix, and what does it actually sell?
It makes memory chips — DRAM and NAND flash — and almost nothing else. The company’s own annual report states that semiconductors exceed 90% of total revenue, which is why it reports as a single segment rather than breaking out divisions (rcpNo 20260317000635).
Its main products are DRAM, NAND flash and multi-chip packages. The DRAM matters most: SK hynix is the leading supplier of high-bandwidth memory (HBM), the stacked DRAM that sits beside AI accelerators and is currently the scarcest input in the AI build-out. It also owns Solidigm, the former Intel NAND business, as a consolidated subsidiary.
Headquarters are in Icheon, Gyeonggi Province, about an hour southeast of Seoul. The chief executive is Kwak Noh-jung. It is listed on the KOSPI as 000660 and, since July 2026, on Nasdaq as SKHY. Further background: Wikipedia.
🌍 Is it actually a global leader?
In DRAM, yes — it reports roughly a third of the world market. The annual report carries market-share figures sourced to IDC:
| Market share (revenue basis) | 2023 | 2024 | 2025 Q1 | 2025 Q2 | 2025 Q3 |
|---|---|---|---|---|---|
| DRAM | 29.9% | 33.4% | 36.3% | 39.1% | 34.3% |
| NAND flash | 19.6% | 21.6% | 16.1% | 22.0% | 19.7% |
As disclosed in the annual report, sourced by the company to IDC (DRAM Nov 2025, NAND Dec 2025), revenue basis. These are the most recent quarters the report carries.
The NAND line is the volatile one — 16.1% to 22.0% inside a single year — which matters for the section on why this stock trades like a NAND company.
🗓️ How SK hynix got here
It started as a construction company. The corporate history in the annual report begins in a place most English coverage never mentions:
| Oct 1949 | Founded as Kukdo Construction |
| Feb 1983 | Renamed Hyundai Electronics Industries — the entry into semiconductors |
| Mar 2001 | Renamed Hynix Semiconductor after the Hyundai group’s restructuring |
| Mar 2012 | Acquired by SK Group; becomes SK hynix |
| Oct 2021 | Develops HBM3, an industry first — the product behind the current cycle |
| Dec 2021 | Agrees to acquire Intel’s NAND business (later Solidigm) |
| Apr 2023 | First 12-layer HBM3; issues the $1.7bn exchangeable bond that still shapes the share count |
| Jul 2026 | Nasdaq listing via a $26.5bn new-share issue |
Corporate history and product milestones from the annual report’s 회사의 연혁 section (rcpNo 20260317000635).
Two entries on that list do most of the work in this article. The 2023 exchangeable bond is why the share count rose even after a cancellation, and the 2021 Intel deal is why a DRAM company carries NAND exposure on its balance sheet as well as in its fabs.
📊 What the Company Reported
All figures below are from the Q2 earnings filing (rcpNo 20260729800013, 29 July 2026) and the FY2025 audited statements.
| 2026 Q2 | 2026 Q1 | 2025 Q2 | 2025 FY | |
|---|---|---|---|---|
| Revenue | $56.0bn | $37.1bn | $15.7bn | $68.6bn |
| Operating profit | $42.8bn | $26.6bn | $6.5bn | $33.3bn |
| Pre-tax income | $86.7bn | $36.5bn | $6.2bn | $35.6bn |
| Net income | $66.3bn | $28.5bn | $5.0bn | $30.3bn |
| Non-operating (pre-tax − operating) | $43.9bn | $9.9bn | −$0.3bn | $2.3bn |
In won: Q2 revenue ₩79,318,746m · operating ₩60,542,608m · pre-tax ₩122,708,355m · net ₩93,922,593m. The non-operating row is not a disclosed line; it is pre-tax minus operating, both reported. $1 = ₩1,416.10, the 7 August 2026 close (Seoul Money Brokerage Services), used throughout except where a filing states its own rate.
🤔 Why is SK hynix’s non-operating income larger than its operating profit?
Because roughly half of the quarter’s pre-tax income did not come from selling memory. Q2 non-operating income of $43.9bn (₩62.2tn / $43.9bn) exceeded operating profit of $42.8bn (₩60.5tn). The equivalent line was ₩3.3tn for the whole of 2025. Sequentially the move is smaller but still steep: ₩14.0tn ($9.9bn) in Q1 to ₩62.2tn ($43.9bn) in Q2, 4.4× in one quarter. Either way it is the largest single fact in these accounts.
The half-year report is not published yet, so the Q2 notes are unavailable. What can be done is a reconciliation against the Q1 report (rcpNo 20260515002287), and one candidate fits the size closely.
🔍 The KIOXIA Reconciliation — and What Is Inferred
SK hynix holds a stake in KIOXIA, the Japanese NAND maker, acquired through the Bain-led consortium that bought Toshiba Memory. KIOXIA trades in Tokyo (285A.T) and rose 370.0% over the second quarter.
The arithmetic. Q1 long-term investment assets were ₩20.66tn ($14.6bn, disclosed). If X is the KIOXIA portion:
X × (¥89,680 / ¥19,080 − 1) = X × 3.700 = ₩62.2tn → X ≈ ₩16.8tn ($11.9bn)
That implies KIOXIA was about 81% of long-term investment assets. It is by far the largest such holding, so the share is plausible — but 81% is our inference, not a disclosure.
Three things could make it wrong. The stake is held indirectly through BCPE Pangea, so the carrying value may be a limited-partnership fair value rather than the listed price. Part of the position was being realised: Q1 cash flow shows ₩4.12tn of long-term investment asset disposals alongside ₩3.95tn of dividends received. And any other holding could account for some of the move. The half-year report’s investment-assets and related-party notes settle all three, and it is due within weeks.
What happens to an unrealised mark when the asset falls?
It reverses — to the extent the position is still held. Since 30 June, KIOXIA is down 46.8% and the NAND complex with it.
| Close | 31 Mar 2026 | 30 Jun 2026 | Change to 7 Aug |
|---|---|---|---|
| KIOXIA (285A.T) | ¥19,080 | ¥89,680 | −46.8% |
| SanDisk (SNDK) | $635 | $2,274 | −46.7% |
| Micron (MU) | $338 | $1,154 | −24.0% |
| SK hynix (000660.KS) | ₩807,000 | ₩2,650,000 | −46.3% |
SK hynix from Korea Exchange settled closes. Peer closes are public market data as of 7 August 2026 and we do not hold them as primary sources; treat the percentages as indicative to the nearest point.
If the Q2 mark was what the arithmetic suggests, the quarter-end carrying value was around ₩79tn ($55.8bn), and a 46.8% fall implies a reversal in the order of ₩37tn ($26.1bn) in Q3. That is an inference resting on an inference — and the Q1 disposals above cut directly against its size. What does not depend on the estimate is the direction: a large unrealised gain becomes a large unrealised loss unless the position was sold. Operating profit could hold near ₩60tn in Q3 and pre-tax income still fall by more than half.
💰 What This Does to the Multiple
First-half cumulative: revenue ₩131.9tn ($93.1bn), operating profit ₩98.2tn ($69.3bn), pre-tax ₩174.3tn ($123.1bn), net income ₩134.3tn ($94.8bn). At the 7 August 2026 close of $1,004 (₩1,422,000) and 730,492,365 shares — market capitalisation $733.5bn:
| Basis | Earnings | EPS | P/E |
|---|---|---|---|
| Annualised H1 net income | $189.5bn | ₩367,275 | 3.9× |
| Annualised operating profit, taxed at 26.4% | $102.0bn | ₩197,882 | 7.2× |
| 2025 actual (disclosed EPS ₩62,044) | $30.3bn | ₩62,044 | 22.9× |
3.9× and 7.2× are both honest. They answer different questions. The first asks what the company earned; the second asks what the memory business earned. When much of the gap is an unrealised mark that has already reversed in the market, the second is the better anchor. 7.2× is not expensive at this point in a cycle — but it is not 3.9×.
26.4% is Korea’s top corporate rate including the local surtax. Q2’s actual effective rate was 23.5%, which would give a slightly lower multiple. Annualising a half-year at a cyclical peak flatters any memory company; we show it because the market quotes it, not because we lean on it. EPS uses net income attributable to owners of the parent (H1 ₩134,150,412m). The first two rows divide by the 7 August share count of 730,492,365; the trailing row uses the EPS the company disclosed for FY2025, which rests on that year’s weighted-average share count. Dividing the 7 August market capitalisation by FY2025 net income instead gives 24.2×.
📉 Why does SK hynix trade like a NAND company?
Partly because it is one. This is the point where our first draft of this article was wrong, and the correction matters enough to state plainly.
SK hynix bought Intel’s NAND business and operates it as Solidigm, a consolidated overseas subsidiary — confirmed in the company’s own filing of 5 August 2026 (rcpNo 20260805800806), which responds to a press report about a pre-IPO for the unit and opens: “당사의 해외 종속회사인 솔리다임은…” — “Solidigm, the company’s overseas subsidiary.” That establishes the ownership; the filing says nothing about production, and SK hynix does not break out the unit’s results. What it does mean is that some of the group’s NAND exposure sits inside consolidated operations rather than in a shareholding.
| Since 30 June | Change | NAND exposure |
|---|---|---|
| Micron | −24.0% | DRAM-weighted |
| SanDisk | −46.7% | Pure NAND; KIOXIA joint-venture partner |
| KIOXIA | −46.8% | Pure NAND |
| SK hynix | −46.3% | Solidigm (owned fabs) and the KIOXIA stake |
The three names clustered at −46 to −47% all carry NAND. Micron, the DRAM-weighted one, fell 24%. We are not going to tell you how much of the 22-point gap belongs to the stake and how much to Solidigm — the company does not disclose Solidigm’s contribution separately, and without that the split is unknowable from filings. Both channels point the same way, which is the honest limit of what this data supports.
🏛️ Does weak shareholder return explain the underperformance?
No — but not because returns improved. The company built the capacity to pay more and has so far used 13.2% of it. This sequence is only visible in Korean filings, and it cuts against both the popular story and our own first reading of it.
At the annual meeting on 25 March 2026, shareholders approved agenda item 9, 자본준비금 감소의 건 — a reduction of capital surplus — with 99.9% of votes cast (77.7% of shares outstanding; rcpNo 20260325801452). The convocation notice (rcpNo 20260305001231) states the purpose in the company’s own words:
“주주환원 효과 극대화를 위한 재원 확보를 위해 회사의 자본준비금 87,172억원 중 40,836억원을 감액하여 이익잉여금으로 전입”
“To secure the resources to maximise the shareholder-return effect, ₩4,083.6bn of the company’s ₩8,717.2bn capital surplus (separate-basis) is reduced and transferred to retained earnings.” — under Article 461-2 of the Commercial Act.
$2.88bn of capital surplus became distributable profit. Korean companies cannot pay dividends out of capital surplus, and unrealised valuation gains do not count towards distributable income either — so a company booking ₩62.2tn of unrealised gains does not gain ₩62.2tn of dividend capacity. We cannot show the constraint was binding — FY2025’s ₩2.1tn of dividends came out of ₩42.9tn of net income — only that the board chose to widen the headroom before the record quarter. The board addressed that in March, before the record quarter.
What has it done with the capacity?
So far, kept paying the same dividend.
| Distributable profit created, March 2026 | $2.88bn (₩4,083.6bn) |
| Declared since — 22 April, record date 31 May | ₩375/share · $187.7M |
| Declared since — 7 August, record date 31 August | ₩375/share · $193.0M |
| Share of the March headroom distributed | 13.2% |
The denominator here is the March headroom specifically, not the company’s whole distributable pool — consolidated retained earnings were ₩106.6tn at end-2025, so the ₩4,083.6bn transfer was a 3.8% increment to it.
₩375 is not a new rate. FY2025’s ₩3,000 per share was ₩1,875 year-end plus ₩1,125 of interim and quarterly payments — three payments of ₩375. The quarterly dividend has been unchanged through 2025 and 2026. The 7 August declaration (rcpNo 20260807800589) carries a stated yield of 0.02% and amounts to 0.29% of the quarter’s net income — a denominator itself inflated by the non-distributable unrealised gain described above. The cleaner anchor is the company’s own disclosed FY2025 payout ratio: 4.90%, on total dividends of $1.48bn — from DART’s annual dividend disclosure, not the quarterly filing linked above.
Two further items from the same meeting the draft would be wrong to leave out. Agenda item 11, 2026년 자기주식 보유 및 처분 계획 승인의 건 — a plan to hold and dispose of treasury stock, not cancel it — passed with 88.6% in favour. Approval of 88.6% was the second-lowest of the meeting’s fifteen voted line items; only the re-election of one outside director drew more opposition, at 83.2% in favour. The remainder cleared between 92% and 99.9%. And FY2025’s own disclosure puts the dividend yield including interim payments at 0.3%, against the 0.40% carried in the annual-report field; the company reports both, on different bases.
Is anything actually changing?
The company has committed to a date. The 7 August filing closes with a sentence that most coverage of the dividend skipped:
“추가 주주환원 방안을 적극 검토 중이며, 구체적인 사항은 3분기 중 확정하여 발표 예정”
“Additional shareholder-return measures are under active review; the specifics will be finalised and announced during the third quarter.”
Read against the rest of the year, that sentence sits at the end of a sequence: capacity created in March, a treasury plan that drew the meeting’s second-heaviest opposition, an unchanged dividend through two quarters, and now a commitment to decide by the end of Q3. We are not going to tell you what motivated it — the filings state actions and dates, not intent.
One more absence is worth recording, because it is checkable. Korea’s Value-Up programme has its own disclosure vehicle, 기업가치제고계획, and companies file it voluntarily — Celltrion cancelled 4% of itself under that banner, Yuhan retired ₩425.3bn of stock outright, Woori filed a buy-and-cancel package on the worst day of the crash. Across 767 SK hynix filings since January 2025, there is no Value-Up plan at all. Whatever arrives in Q3 would be the first.
So the honest reading is not “weak by design,” but not “changed” either. On the disclosed numbers returns are thin — FY2025’s payout ratio was 4.90% on ₩2,095,133m of dividends. The machinery was built in March and has been drawn on for about an eighth of its value; the rate has not moved; the treasury plan on the table is disposal rather than cancellation; and the company has put a quarter-end date on deciding what to do about it. That is a falsifiable thing to watch, which is more than either of the stories in circulation offers.
Dividend figures from DART’s alotMatter disclosure and the individual dividend filings. Data-vendor yield fields for these tickers return values several multiples away from the disclosed ones and should not be used.
💵 The Nasdaq listing issued new shares
We covered the listing when it happened. The offering filing (rcpNo 20260715000004) adds what most coverage did not: this was not a sale of existing stock.
| New shares issued | 17,790,000 — third-party allotment, +2.50% of shares outstanding |
| Total raised | $26,507,100,000 (₩39.89tn) |
| Price | $149 per ADR; 10 ADRs = 1 common share; ₩2,242,301 per share |
| Use of proceeds | ₩39.89tn ($28.2bn) — 100% to facilities. Working capital, debt repayment and acquisitions all nil |
| Depositary · payment date | Citibank, N.A. · 14 July 2026 |
USD terms are the offering’s own; we do not re-convert them at a different rate. A second figure circulates — ₩40,023,070,290,000 — which is the registered issue amount struck at the 9 July pricing-date rate of ₩1,509.90, against the actual proceeds struck at the 14 July payment-date rate of ₩1,504.90. The ₩132.5bn difference is entirely a five-day currency move: 40,023.07 ÷ 39,890.53 = 1.00332, and 1,509.90 ÷ 1,504.90 = 1.00332.
Both readings hold. Against: 2.50% dilution, priced near the high, and buyers of the new ADRs are about 33% underwater in dollars — they paid $149 per ADR and the share is worth roughly $100 per ADR at parity to the 7 August close (the 36.6% figure often quoted is the won loss on the underlying; the difference is the won strengthening from ₩1,504.90 to ₩1,416.10), while SK Square’s holding fell from 20.10% to 20.01% on the group’s own filing (rcpNo 20260805000440, as of 30 July). The change is almost entirely the dilution: on a common-share basis the reporting group’s holding actually rose by 23,369 shares, and the −134,968 figure some sources quote is measured against its previous filing in November 2021 — relevant to the holding company’s NAV discount. For: ₩39.89tn of permanent capital at no interest cost, committed entirely to capacity. A board funding fabs with equity is expressing a view about demand.
The domestic listing of the new shares completed on 29 July, and the share count steps from 712,702,365 to 730,492,365 on that date in exchange data. The stock then fell 14.7% between 5 and 7 August. We are not going to tell you those are connected. The same window contains a 14.65% fall on 28 July — the day before the listing — and a 29.95% limit-up on 31 July. A month that volatile will place almost any date next to almost any move, and the KIOXIA reversal above has better arithmetic behind it. Our flow analysis of that same week found foreign investors net sellers of $2.44bn of SK hynix over five sessions. That figure is struck at that issue’s rate of ₩1,423.51 and we do not restate it here; at this article’s rate the same flow is $2.45bn. The direction holds across windows — net selling of $2.45bn, $3.06bn and $3.74bn over 5, 10 and 20 sessions — so it is not a one-week artifact.
🔄 They cancelled shares — and the float still grew
This is visible only in the share-count series, not in disclosure titles.
| End 2024 | End 2025 | 2026 Q1 | |
|---|---|---|---|
| Shares issued | 728,002,365 | 728,002,365 | 712,702,365 |
| Treasury | 38,963,634 | 26,310,845 | 4,405,344 |
| Free float | 689,038,731 | 701,691,520 | 708,297,021 |
Shares issued fell 2.10% between end-2025 and Q1 — 15,300,000 shares were retired, a genuine cancellation. Over the same window free float still rose 0.94%, because treasury stock left the balance sheet faster than it was cancelled — and by 7 August, after the July share issue, float stands at 728,866,138, +3.87% against end-2025:
Treasury 26,310,845 → 4,405,344 = −21,905,501
cancelled −15,300,000 (accretive)
released −6,605,501 (dilutive)
The released portion is almost entirely a 2023 exchangeable bond: $1.7bn at 1.75%, exchangeable at ₩108,811 — 7.7% of today’s price. Of the 6,605,501 treasury shares released in Q1, 6,179,194 (93.6%) went to bondholders and 426,307 (6.4%) to employee compensation.
A reconciliation note: cumulative exchanges of 17,807,022 shares plus a residual 2,753,353 sum to 20,560,375, against the 20,139,289 shares the Q1 report records as remaining exchangeable at issue — a difference of 421,086. An over-deposit to cover the bond’s exchange-price adjustment is the ordinary explanation, and the instrument ended in a clean-up call with 73 odd-lot shares reacquired in May 2026. It is 0.06% of the float and moves nothing here.
The cheap money in 2023 was not free. It was priced in equity at a memory-cycle trough, and the bill arrived as float in 2026.
⚠️ What would prove this read wrong
- The half-year report’s investment-assets notes. If Q2’s non-operating income is not principally a KIOXIA revaluation, the central inference fails outright. One document settles it, and it is due within weeks.
- If the stake is carried at a limited-partnership fair value rather than the listed price, both the Q2 gain and the implied Q3 reversal shrink, possibly a great deal.
- If the Q1 disposals continued through Q2 at scale, the reversal is far smaller than ₩37tn. The Q1 cash flow already shows ₩4.12tn of disposals; we do not know the Q2 figure.
- If Q3 pre-tax income lands near Q2’s despite KIOXIA’s fall, the mark-to-market mechanism described here is not what drives these accounts.
- If Solidigm’s disclosed contribution turns out to be small, the operational-NAND explanation weakens and more of the peer gap returns to the stake — the opposite of the correction we made above.
📚 Lingo Check
- 영업외이익 (yeong-eop-oe i-ik) — non-operating income: everything between operating profit and pre-tax income. Korean filings disclose both ends, so the line is recoverable by subtraction even before the notes appear.
- 자본준비금 감소 (ja-bon-jun-bi-geum gam-so) — reduction of capital reserves. Under Article 461-2 of the Commercial Act a company may move reserves into retained earnings by shareholder vote, creating distributable profit. It is a governance signal that rarely reaches English coverage.
- 이익소각 (i-ik so-gak) — cancellation of treasury shares out of distributable profit. It permanently reduces shares issued, unlike a disposal, which returns stock to the float.
- 교환사채 (gyo-hwan sa-chae) — exchangeable bond, convertible into shares the issuer already holds in treasury. It dilutes the float without changing shares issued.
📌 A note on our earlier coverage
We published an HBM4E outlook on SK hynix earlier in this cycle which carried an undated Buy (Strong) rating with a target of ₩3,200,000 (stated as ₩2,950,000 elsewhere in the same piece — it contradicts itself) and described a 7.0× forward multiple as an exceptional entry point, struck when the shares were ₩2,425,000. They closed at ₩1,422,000 on 7 August — 41% lower. This article supersedes that valuation framing. The 7.2× figure here is calculated on operating profit only and on the post-issue share count, and it sits beside a Q3 reversal risk that the earlier piece did not contemplate. Our other SK hynix coverage — the supercycle overview, ADR-versus-home-market decoupling and how the USD prices are quoted — stands.
🎯 Why it matters
The question people ask is why a company printing record profits lost half its market value. The answers in circulation are either unfalsifiable or contradicted by the data — the shareholder-return explanation needs narrowing rather than discarding: returns are thin on the filings — a 4.9% payout ratio for FY2025 — but the claim that the company is structurally unwilling does not hold, because it created the headroom in March and has dated a decision to Q3.
What the filings support is duller and more checkable: a large share of the record sits in a non-operating line, the most likely source of that line has since fallen by half, and the company carries NAND exposure through both a subsidiary and a shareholding. The first of those is arithmetic. The second is our inference, and we have named the document that will confirm or kill it.
We would rather publish an inference with its falsification condition attached than an assertion without one.
Sources and method. Q2 results: DART rcpNo 20260729800013. Q1 report and cash flow: 20260515002287. ADR offering: 20260715000004. AGM result: 20260325801452. Quarterly dividend: 20260807800589. Solidigm subsidiary status: 20260805800806. Share counts, treasury and dividend history via DART’s stockTotqySttus, fnlttSinglAcntAll and alotMatter endpoints. SK hynix prices and market capitalisation from the Korea Exchange Open API, settled closes. Peer prices are public market data as of 7 August 2026 and are not primary sources for us. $1 = ₩1,416.10, the 7 August close. The half-year report and its notes were unavailable at the time of writing; every inference drawn in their absence is labelled above.
We log corrections rather than editing silently — see our corrections policy. Research and data, not investment advice; see our investment disclaimer.
Written by James Ju — a Seoul-based engineer who reads Korean-language DART and KRX filings, the primary sources most English-language coverage skips. About the author.