213420 — Duk San Neolux Co.,Ltd

Duk San Neolux (213420.KQ, KOSDAQ) — deep-value analysis

Date: September 18, 2026 · Price: … · Market cap: … · Shares outstanding: 24,562,915 (issued 24,831,179 − treasury 268,264) Primary sources: DART half-year report 20260813001263 (H1 2026), DART annual reports FY2025 and FY2023 (fnlttSinglAcntAll, CFS), DART event reports from 2026-03-26, 2026-05-27, 2026-08-27, 2026-09-02 and 2026-09-15, the FY2025 consolidated statements of Duk San Hi-Metal (077360), data pack data-pack-213420.KQ-20260918.md, research brief research-213420-20260918.md.

Method note. The data pack generated on the tracker symbol 213420 came out EMPTY (yfinance can’t resolve the symbol without a suffix). I regenerated it on 213420.KQ. Even so, yfinance’s balance sheet doesn’t contain current assets (receivables, inventory, detailed cash) for this issuer — every balance-sheet figure on receivables, inventory, loans, and liabilities in this report comes from DART filings, not the data pack. I flag them as such.

Executive summary (1 page: thesis, estimated value, verdict)

Duk San Neolux is, essentially, two companies glued into one stock ticker, plus a third that’s attaching right now. The first is the legacy business: organic materials for OLED displays — hole transport layer (HTL), host, prime layer, and Black PDL, the non-emissive material developed as a world first in 2021. It’s a business with a 24% operating margin, R&D of 10… of revenue, and an anchor client that is the world’s largest maker of small and medium OLED panels. The second is Hyundai Heavy Industries Turbo Machinery, 59.69% bought in February 2025 — large pumps and compressors, 17.2% operating margin in H1 2026, clients like Korea Hydro & Nuclear Power. The third is S&S Valve, an industrial valve maker bought 100% for KRW 144.50 bn, with payment completed on September 3, 2026 — that is, two weeks ago, after the date of the last published balance sheet.

On the reported figures, the stock looks cheap and consensus categorically agrees: trailing P/E 10.1x on attributable net profit of KRW 71.31 bn over the last four quarters, P/B 1.47x, simple FCF yield 9.7%, and the tracker’s 9-analyst average target is … — i.e. … versus the price. The price sits at a three-year low (… on a 3-year index), RSI 40.7, below EMA50 and below EMA200. The pattern looks exactly like a good company sold off in a panic.

Analysis of the primary filings says something else, and the difference isn’t a matter of nuance. Three facts change the calculation, and none of them appear in the research brief, the data pack, or the consensus note.

One: KRW 116.80 bn — 16.3% of the market cap — is lent to the controlling shareholder and an affiliate. Note 28 of the half-year report shows KRW 101.20 bn lent to Duk San Hi-Metal (holder of 36.7% of the company) and KRW 15.60 bn to Duk San Holdings. The 101.20 bn balance is, at the latest figure, unchanged between 2025-06-30 and 2026-06-30 — nothing was amortized in twelve months, and it originates from a … outflow in the 2023 investing cash flow. The collateral is a pledge on 524,800 shares of an unlisted company (Sirius Holdings) plus a third-ranking pledge on 12.5 million Ether City shares. There is no impairment provision. And the debtor, Duk San Hi-Metal, reported in FY2025 a consolidated net loss of KRW 115.70 bn (versus +31.30 bn in FY2024), with operating profit collapsing from 18.64 to 2.81 bn and equity attributable to shareholders falling from 313.38 to …. The loan represents 49% of the debtor’s own equity, and the debtor has KRW 66.55 bn of cash.

Two: the fortress balance sheet disappears in the current quarter. At 2026-06-30 the group had KRW 165.54 bn cash against KRW 135.72 bn total debt (bank 62.45 + convertible bond 20.68 + redeemable preferred shares 40.00 + leasing 12.59). On September 2 the company injected KRW 71.45 bn into the Canopus Holdings vehicle, and on September 3 Canopus paid KRW 144.50 bn for S&S Valve. On November 30, 2026 the convertible bond matures: 20.00 bn face value × 105.10% = … in cash, since the conversion price is … and the stock sits at … — nothing converts. The net financial position moves from +38.4 bn net cash to roughly −106 bn net debt, without even counting the target’s own debt.

Three: inventory has exploded, while the impairment provision was reduced at the same time. Consolidated inventory grew from 65.77 to … in six months (…), with finished goods almost doubling (15.20 → 29.60 bn, …). The valuation provision fell from 8.07 to 6.75 bn, i.e. from 10.9% to 6.0% of gross inventory, and KRW 1.33 bn of impairment reversal was credited directly to cost of sales in H1 2026. Days of inventory (DIO) climbed from 105 to 153.

My valuation, with pro-forma normalized earnings of …, growth 7%, discount …, terminal growth 2%, and pro-forma net debt of … (after crediting the affiliate loans at a 40% recovery), gives a range from … (EPV Greenwald) to … (DCF bull), with the median of the five models at … and the Monte Carlo median over 20,000 scenarios at …, with an undervaluation probability of …%. In other words: at … the stock isn’t expensive, but it doesn’t offer the margin of safety that the 10x P/E and the … consensus target suggest. The apparent discount is largely the correct discount for KRW 4,755/share locked in a loan to a shareholder that’s losing money, for a KRW 215.5 bn diversification into industries unrelated to OLED, and for zero dividends in the company’s entire history.

Verdict: MONITOR. Convergent with the prior GBL score in the tracker (61.7%, “MONITOR,” evaluated 2026-09-06), but for completely different reasons — see the last chapter. Central intrinsic value: 28,800-…/share, with a defensible range of 19,500-50,500. A buy is justified below …, where the pessimistic range becomes covered.