SK Hynix (000660) 🔎 In-depth
KOSPI · Price as of 2026-11-12 · Metrics 2026-11-13 · In-depth research 2026-11-07
SK Hynix makes memory chips such as DRAM and NAND flash. High-bandwidth memory (HBM), which sits next to the compute chips in AI servers, is the core of its profit, and virtually all of its revenue comes from memory. In Q3 2026, quarterly revenue topped ₩100 trillion for the first time at ₩104.6155 trillion, with an operating margin of 79.7%, and the company put its expected shareholder returns for this year at about ₩100 trillion. The key point now is two-sided. While long-term supply agreements with 14 customers and net cash of ₩98.66 trillion underpin its results and financial strength, the share price is low relative to this year's expected earnings. If memory price gains keep narrowing, however, or a strong won and valuation losses on investment assets hit at the same time, the earnings figures could wobble as well.
This page organises public market and filing data for information purposes. It is not a recommendation to buy or sell, nor investment advice. Metrics labelled “forward” are our own unverified estimates. Please verify with the original DART filings and decide at your own responsibility.
Data sources
Prices & market cap from the FSC securities price API (data.go.kr) · filings & financials from OpenDART (FSS) · listings from the Korea Exchange.
Prices as of 2026-11-12. Methodology is public on the methodology page.
30-second brief
Auto-extracted from the computed values.
What do the SourceComputedEstimateReading tags mean?
Source filings and exchange data · Computed calculated from public data · Estimate our own unverified projection · Reading drafted automatically, then rule-checked. See AI & Automation.
Core valuation metric
This stock's effective sub-sector is “Memory Chips” (Semiconductors & IT Components · Semiconductors), a type typically read first through P/B.
Memory chipmakers ride the DRAM and NAND price cycle, so profits balloon in upturns and can swing to losses in downturns. That makes earnings-based multiples unreliable, so price-to-book (P/B) — the share price against the company's asset value — is the first lens.
Price against assets alone says little about where the cycle stands. Reading it together with price against this year's expected earnings shows how far profits have recovered.
This note explains how each sector is typically valued and is educational context only, not investment advice.
At-a-glance assessment financial health · growth · profitability · valuation
- Debt ratio, current ratio and interest burden all look healthy.
- Revenue rose 46.8% year over year, and the pace is slowing (3-year trend: rising).
- Most recent quarter (Q1 2026) revenue was 198.1% higher than a year earlier.
- ROE is 45.8% (controlling-interest basis). It is above the sector average.
- Operating margin is 58.6%.
- A forward P/E of 3.77x puts the price low relative to earnings.
Ownership & governance As of 2025-12-31
Largest shareholder SK Square 20.07% (corporate)
Controlling bloc incl. related parties 20.07%
With the controlling bloc holding 20%, control is maintained but the free float is relatively large.
🔎 In-depth analysis Reading
SK Hynix makes and supplies memory chips. Memory comes in two main types: DRAM, which moves data quickly while the power is on, and NAND flash, storage memory that keeps data even when the power is off. In Q3 2026, DRAM made up 75% of revenue and NAND 24% (including U.S. subsidiary Solidigm, per the company's IR materials). In other words, virtually all revenue and profit come from memory. The core of its profit is high-bandwidth memory (HBM), which stacks DRAM in multiple layers. Because it sits right next to the compute chip (GPU) in an AI server and moves large amounts of data at once, it is priced far above standard DRAM. The company said its next-generation HBM4 will surpass the previous-generation HBM3E in both shipments and revenue in Q4. Its key customers are the large cloud companies building AI data centers. The company said it has signed long-term supply agreements with 14 major customers, including these, and that the secured volumes cover more than 50% of its total capacity for 2027 and 2028. The contracts typically run five years, with prices reflecting market conditions within a range agreed with each customer. The PC and smartphone side, by contrast, is weak. In its half-year filing, the company said it expected PC shipment adjustments and some softening in demand in 2026, as higher memory prices raised procurement costs for device makers. Production is centered on Icheon in Gyeonggi Province and Cheongju in North Chungcheong Province, and the company also runs plants in Wuxi (DRAM) and Dalian (NAND) in China. Cheongju's M15X is expected to reach full capacity at the end of 2026. Separately from its operations, the company holds an indirect investment in Japan's Kioxia; this asset is marked to market value and swings net profit sharply every quarter. The largest shareholder is SK Square, with a 20.00% stake.
The latest close is ₩1,254,000 and the market capitalization is ₩916.0 trillion. The price sits below its 20-day moving average (₩1,307,700) and below its 60-day moving average (₩1,572,150). It is under both its short- and medium-term moving averages, so the trend looks subdued. The RSI (a supplementary indicator that gauges the strength of gains versus losses over the past 14 days on a 0-100 scale) is 42.0, a neutral level. The one-month change is -11.6%, the three-month change is -11.7%, and the position relative to the 52-week high is -57.0%. Relative strength versus the KOSPI is 74 (on a 1-99 scale, converted from returns against the index over the past year with more weight on recent performance; higher means stronger than the market). It is stronger than roughly 75% of all stocks. Over the past three months it outpaced the index by 3.9%. Chart interpretation is best done alongside trading volume and the dates on which disclosures occur.
Updating the figures to the latest quarter shows a share price that is low relative to earnings and a very strong balance sheet. First, the denominators behind the default on-screen metrics are out of date. The P/E (how many times one year's earnings the share price is) of 20.99x is based on 2025 full-year earnings, and the P/B (how many times the company's net assets the share price is) of 5.62x is based on equity at the end of Q1 2026. Since then, tens of trillions of won in profit have been added to equity every quarter, so multiples recalculated on current figures are much lower. In the memory industry, earnings swing widely with the price cycle, so P/B is the first metric to check. Recalculated using Q3-end total equity of ₩309.088 trillion (unaudited, including non-controlling interests), the P/B is 5.58x, about half the on-screen figure. On a free-float basis excluding treasury shares, it is 3.19x. ROE (how much the company earns on its equity in a year) is 61.1% over the last four quarters. Taken together, these two figures mean the company earns profit equal to about 18.5% of its share price each year at the current price. The P/E on the last four quarters' earnings is 5.41x, and the forward P/E on this year's expected earnings is 3.80x (an in-house estimate). Profitability is the highest in the company's history. The Q3 operating margin was 79.7%, and 74.5% over the last four quarters. Net profit, however, moves separately from operating results. In Q3, valuation losses on investment assets of ₩22.92 trillion and foreign-exchange-related losses of ₩11.63 trillion from the stronger won pushed non-operating items to −₩32.6253 trillion. As a result, net profit of ₩39.7756 trillion was less than half of operating profit of ₩83.3338 trillion. In the first half, by contrast, valuation gains on investment assets boosted net profit. Keep in mind that P/E and ROE calculated from net profit include these non-operating items. Financial strength is excellent. At the end of Q3, cash and equivalents were ₩114.46 trillion and borrowings ₩15.8 trillion. Cash exceeds debt by ₩98.66 trillion, a net cash position (about ₩69.4 trillion at the end of Q2), and the debt ratio (liabilities relative to equity) is 35.6%. Because of the net cash, EV (market cap plus net debt, the value of the whole company after accounting for debt and cash) is about ₩922.57 trillion, less than the market cap. EV/EBIT (how many times operating profit the whole company is valued at) is 4.60x on the last four quarters and about 3.3x on this year's expected operating profit. EV/EBITDA (the multiple of operating profit before depreciation) is 4.26x, and EV/Sales (the multiple of revenue) is 3.43x. Free cash flow over the last four quarters (cash from operations minus capital spending) was about ₩123.5 trillion. The FCF yield (free cash flow as a percentage of market cap) is about 12.1%, and about 10.7% even after excluding ₩13.9836 trillion in non-recurring dividends received in cash in the first half. The on-screen EV metrics (EV/EBIT of 13.04x, FCF yield of 2.43%) are based on the end-2025 balance sheet and cash flows and on 12-month operating profit through Q1, so they miss these changes. Large outflows also lie ahead: the Q3 dividend of ₩14.1035 trillion payable in December, the ₩20.0077 trillion treasury share acquisition approved in late October, and 2027 capital spending, which the company has said will rise 'significantly'. Per the company's IR materials, its credit grades are AA+ from Korea's three domestic credit agencies, A3 from Moody's and BBB+ (positive outlook) from S&P.
Memory is a business that rides big price cycles, so it makes sense to start with the five-year picture. Annual revenue was ₩43 trillion in 2021, ₩44.6 trillion in 2022, ₩32.8 trillion in 2023, ₩66.2 trillion in 2024 and ₩97.1 trillion in 2025. Operating profit went from ₩12.4 trillion in 2021 and ₩6.8 trillion in 2022 to a loss of ₩7.7 trillion in 2023, then recovered to ₩23.5 trillion in 2024 and ₩47.2 trillion in 2025. Swings this wide between good and bad years are a basic trait of the company. In 2026 it stepped up another level from there. Quarterly revenue rose every quarter, from ₩52.5763 trillion in Q1 to ₩79.3187 trillion in Q2 and ₩104.6155 trillion in Q3 (preliminary), crossing ₩100 trillion for the first time in Q3. Operating profit was ₩37.6103 trillion in Q1, ₩60.5426 trillion in Q2 and ₩83.3338 trillion in Q3. Cumulative Q1-Q3 operating profit of ₩181.4867 trillion is about 3.84 times 2025 full-year operating profit. Because seasonality and the price cycle work together in memory, the standard comparison is with the same quarter a year earlier rather than the previous quarter. Q3 revenue rose 327.9% and operating profit 632.1% from a year earlier. Since Q3 operating profit a year ago was much smaller at ₩11.3834 trillion, bear in mind that these growth rates look large. Q3 growth came from both price and volume. According to the company's IR materials, DRAM shipments rose about 10% from the prior quarter and the average price rose in the high 20s percent. NAND shipments dipped slightly, but the average price rose in the mid-20s percent. Net profit followed a different path. Q3 net profit of ₩39.7756 trillion fell 57.7% from the prior quarter, not because operations weakened but because of non-operating losses such as valuation losses on investment assets and foreign-exchange-related losses. The company has not given full-year revenue or profit figures for this year, so the estimate here was built in-house by stacking Q4 on top of Q1-Q3 results. The Q4 starting point is the company's stated shipment plan. It plans to raise DRAM shipments by about the mid-20s percent, through expansion of its 1c process (the sixth generation of 10-nanometer-class process technology) and higher output at Cheongju's M15X, and NAND shipments by the mid-single digits. Prices were assumed to keep rising but by less than in Q3, with a larger HBM4 share in Q4 supporting unit prices. On the other hand, the won/dollar rate since October has been lower than the Q3 average, a headwind that shrinks revenue in won terms even at the same dollar prices. On these assumptions, Q4 revenue and operating profit rise from Q3. The forward P/E on this year's expected earnings is 3.80x, and about 4.6x on an operating basis that strips out all non-operating items. About 30% of this year's expected net profit is still the Q4 estimate, so the net profit figure could change with Q4 valuation gains or losses on investment assets and with the exchange rate.
Recent disclosures fall into four groups. The first is Q3 results. On October 28, the company disclosed preliminary quarterly revenue of ₩104.6155 trillion, operating profit of ₩83.3338 trillion and net profit of ₩39.7756 trillion. Revenue topped ₩100 trillion for the first time, and the operating margin was 79.7%. However, valuation losses on investment assets of ₩22.92 trillion and foreign-exchange-related losses of ₩11.63 trillion pushed net profit below the prior quarter. On the same day, the company said it had signed long-term supply agreements with 14 major customers, securing more than 50% of its 2027 and 2028 capacity. In Q4 it plans to raise DRAM shipments by about the mid-20s percent and NAND by the mid-single digits. The second is shareholder returns. The company put its expected shareholder returns for fiscal 2026 at about ₩100 trillion, noting that this is forward-looking information and actual results may vary with business conditions. The plan consists of at least ₩30 trillion in cash dividends and at least ₩60 trillion in treasury share acquisitions and cancellations (including the ₩40 trillion decided in August). The Q3 dividend is ₩20,000 per share, ₩14.1035 trillion in total, with a record date of November 30. That is a large increase from the Q2 dividend (₩375 per share), but ₩10,000 of it is a special dividend distributing gains realized on the Kioxia investment, so it is not the kind of payment that repeats every quarter. The ₩40 trillion treasury share acquisition decided in August was completed on October 21: 23,693,272 shares were acquired at an average of ₩1,688,426, with cancellation scheduled for November 5. Once they are cancelled, shares issued will fall from 730,492,365 to 706,799,093. As of November 7, no filing confirming the cancellation had yet been found. On October 28, the board also approved an additional treasury share acquisition of ₩20.0077 trillion (through January 28, 2027). About ₩10 trillion is to be cancelled and about ₩10 trillion used for employee stock compensation. The third is investment. On July 14, the company raised ₩39.8905 trillion by issuing American depositary receipts (ADRs) in the U.S., all earmarked for capital spending. On August 7, it approved investments of ₩35.2246 trillion in the second fab at the Yongin semiconductor cluster and ₩19.1 trillion in Cheongju M17. An October 6 filing put expected Q4 construction transactions with SK ecoplant at ₩4.2462 trillion, more than 20% above the amount disclosed on February 27. The company said the figure reflects increases from new and changed projects as well as construction advances that arose from paying suppliers in cash instead of promissory notes. The company has signaled that 2027 investment will be significantly larger than in 2026. The fourth is matters still undecided. The pre-IPO fundraising and IPO review for Solidigm were confirmed as halted on October 23 (the company cited capital market conditions, among other factors). In response to reports of a new Solidigm NAND plant in the U.S. (re-disclosure due November 27), a production base in Japan (December 17) and the divestment of a stake in a back-end facility in Chongqing, China (December 8), the company replied in every case that the matter is 'under review and undecided'. Depending on the outcomes, the scale of investment and the mix of production sites could change.
Start with the points to watch. First, customers' investment is still growing. On October 21, Meta narrowed its 2026 capital spending range to $135-145 billion by raising the low end, and said it will keep investing aggressively in AI in 2027. Large cloud companies' 2027 investment budgets will come with their earnings releases in January-February next year. If increases are confirmed, additional orders would stack on top of long-term contract volumes and extend the supply shortage. Second, most of the output from competitors' new fabs arrives only after the second half of 2027. On September 30, Micron indicated first wafer output from its first Idaho fab in mid-2027 and from its first New York fab in 2030. Until then, industry supply growth is limited mainly to process migrations. The company itself expects demand to exceed supply through 2030. Third, a large share of future revenue is locked in by contract. Long-term supply agreements with 14 customers cover more than half of 2027 and 2028 capacity, with performance safeguards such as deposits attached. Fourth, the valuation is low relative to earnings. A forward P/E of 3.77x, an EV/EBIT of about 3.3x on this year's expected operating profit and an FCF yield of about 0.2% are all low levels. On top of net cash of ₩98.66 trillion, this year's shareholder returns (about ₩100 trillion), equal to about 10% of market cap, are also under way. The cautions are just as clear. First, prices are rising, but the pace is slowing. Micron expects price increases to become more gradual. Kioxia also guided October-December revenue and profit to be similar to the prior quarter despite strong data-center demand, citing currency effects among other factors. With the operating margin already close to 80%, once contract-price gains approach flat, profit growth will depend on shipment volumes. If prices start to fall, earnings and multiples would retreat together. The price-range structure of the long-term agreements cushions downturns but caps the upside in upswings. Second, competition is intensifying. Samsung Electronics sharply increased HBM4 shipments in Q3 and said it will further expand HBM and server DRAM supply in Q4. According to the half-year filing, SK Hynix's revenue-based DRAM market share fell from 34.8% in 2025 to 29.1% in Q1 2026. The company itself expects the profitability gap between HBM and standard DRAM to narrow meaningfully in 2027. China's CXMT is also a competitor: in its July U.S. listing prospectus, the company named CXMT alongside Samsung Electronics and Micron as a major DRAM competitor. Third, customers are finding ways to use less memory. The company's half-year filing noted growing optimization of server memory configurations, and Micron also said growth in memory content per server is running below expectations. In March, Google Research released a technique that compresses the memory used for AI inference. The longer memory prices stay high, the faster such memory-saving technologies spread. Conversely, if lower inference costs make AI usage grow faster, overall demand could actually increase. Fourth, policy and exchange rates. The U.S. withdrew the blanket authorization it had granted the Wuxi and Dalian plants in China and replaced it with annual equipment import licenses. The 2027 license must be obtained again by year-end, and delays could disrupt equipment maintenance and replacement. Most revenue is settled in dollars, so a stronger won reduces revenue in won terms and creates foreign-exchange losses. The company said a 10% move in the dollar exchange rate changes pre-tax profit by ₩4.7517 trillion (as of end-June). Fifth, swings in net profit and the burden of investment. The Kioxia-linked investment asset is marked to market and could jolt net profit again in Q4. A 'significant increase' in 2027 investment has been flagged, and the medium- to long-term investment guideline is about ₩1,100 trillion. In sum, as long as cloud companies' higher 2027 investment and competitors' expansion timelines are confirmed, contract volumes and cash strength support earnings and the current low multiples stand out. Conversely, if quarterly contract-price gains stall, if Samsung Electronics' or CXMT's volumes grow faster than expected, or if a strong won and investment-asset valuation losses coincide, the earnings figures and multiples could wobble together.
🔎 Valuation vs peers Undervalued
In Korea, Samsung Electronics is the only direct peer producing DRAM and NAND at scale, and it is only partly comparable because it also runs smartphones, appliances, foundry and displays; Hanmi Semiconductor (HBM back-end equipment) and DB HiTek (8-inch contract manufacturing) belong to the same semiconductor supply chain but have different business structures, so they serve only as references for the multiples different parts of the chip industry receive. Peer figures are comparison-tool values, and Samsung's 26.62x P/E is based on 2025 annual earnings and therefore outdated; recalculated the same way from DART confirmed and preliminary quarters (last four quarters' net profit attributable to controlling shareholders of ₩212.8708 trillion, market cap of ₩1,283.2582 trillion), Samsung's trailing P/E is about 6.0x and SK Hynix's about 5.4x. The default industry median rests on just two samples, so its signal is weak.
| Peer | P/E | P/B | ROE |
|---|---|---|---|
| Samsung Electronics | 27.47x | 2.57x | 17.57% |
| Hanmi Semiconductor | 81.01x | 27.27x | 28.04% |
| DB HiTek | 13.98x | 1.57x | 13.04% |
In short, the stock looks undervalued on this year's expected earnings. First, its position versus peers. Calculated the same way on the last four quarters' earnings as Samsung Electronics, which rides the same memory cycle, SK Hynix's P/E (how many times one year's earnings the share price is) is about 5.4x versus about 6.0x for Samsung, so SK Hynix is lower. The forward P/E of 3.77x on this year's expected earnings is less than half the 10.42x median forward P/E across all stocks on the site. Because the memory-chip industry has only two comparable samples, this median reflects all stocks rather than the industry and is for reference only. Second, the case for a higher and a lower valuation. The shift to HBM4 as the main product, long-term agreements covering more than half of 2027 and 2028 capacity, net cash of ₩98.7 trillion and this year's shareholder returns of about 10% of market cap argue for a higher valuation. Large swings in net profit from investment-asset valuations and exchange rates, the surge in 2027 investment, the cap on long-term contract prices, the narrowing profitability gap between HBM and standard DRAM, and the decline in DRAM market share argue for a lower one. Third, the on-screen P/E of 20.82x is based on 2025 full-year earnings and does not capture the current picture: Q1-Q3 operating profit alone is already 3.84 times last year's full-year figure. The forward P/E of 3.80x, built quarter by quarter (about 4.6x on an operating basis that strips out non-operating items), is a low level. EV/EBIT (how many times operating profit the whole company, including debt and cash, is valued at) is 4.6x on the last four quarters and about 3.3x on this year's expectations, and the FCF yield (free cash flow as a percentage of market cap) is about 12%. P/B (how many times net assets the share price is), which this industry checks first, is 3.30x on Q3-end equity, about half the on-screen value of 6.22x. Given an ROE (how much the company earns on its equity in a year) of 61%, the earnings yield at the current price is about 18.5%. This view, however, depends on the premise that memory prices keep rising in Q4, only by smaller amounts. If prices turn flat or start falling, earnings and multiples could retreat together.
Price history Close · MA20 · MA60
The latest close is ₩1,254,000 and the market capitalization is ₩916.0 trillion. The price sits below its 20-day moving average (₩1,307,700) and below its 60-day moving average (₩1,572,150). It is under both its short- and medium-term moving averages, so the trend looks subdued. The RSI (a supplementary indicator that gauges the strength of gains versus losses over the past 14 days on a 0-100 scale) is 42.0, a neutral level. The one-month change is -11.6%, the three-month change is -11.7%, and the position relative to the 52-week high is -57.0%. Relative strength versus the KOSPI is 74 (on a 1-99 scale, converted from returns against the index over the past year with more weight on recent performance; higher means stronger than the market). It is stronger than roughly 75% of all stocks. Over the past three months it outpaced the index by 3.9%. Chart interpretation is best done alongside trading volume and the dates on which disclosures occur.
Relative performance stock vs index · start = 100
Excess return vs index · 3M +3.88% / 6M -2.83% / 12M +58.40%
Key metrics Computed vs whole-market median
Valuation
The P/E of 20.82x is above the whole-market median (12.03x). The P/B of 5.58x is above the whole-market median (0.77x). That said, this P/E is based on last year's (trailing) results. With recent quarterly earnings up sharply, the trailing P/E can look higher than it really is, so a precise read is best done on this year's expected (forward) earnings.
Enterprise value (EV)
EV = market cap + net debt. It reflects cash and debt, so it captures the real cost of the whole business that market cap alone misses; lower multiples are cheaper relative to earnings or sales.
Intrinsic value (DCF estimate) Estimate
Model output — not a price target. This is what the formula returns under the assumptions below, not a view on where the share price should go.
DCF (discounted cash flow) estimate — discount rate 11.0%, initial growth 10.0%→terminal 2.0%, 10-yr forecast, free-cash-flow basis, forward earnings power normalized 3x. A reference range that shifts materially with assumptions.
Profitability & financials
Return on equity (ROE) is 45.8%, above the whole-market average (3.0%). The operating margin is 58.6%. The debt ratio is 35.6%, so the financial structure is stable.
Growth FY2025 · annual report (consolidated)
| Item | 2023 | 2024 | 2025 | YoY |
|---|---|---|---|---|
| Revenue | $24.6B | $49.6B | $72.8B | +46.76% ↓ slower |
| Operating profit | -$5.8B | $17.6B | $35.4B | +101.16% |
| Net profit | -$6.8B | $14.8B | $32.2B | +116.89% |
| 5-year | 2021 | 2022 | 2023 | 2024 | 2025 |
|---|---|---|---|---|---|
| Revenue | $32.2B | $33.4B | $24.6B | $49.6B | $72.8B |
| Operating profit | $9.3B | $5.1B | -$5.8B | $17.6B | $35.4B |
| Net profit | $7.2B | $1.7B | -$6.8B | $14.8B | $32.2B |
| Revenue CAGR | 4-yr avg 22.60% | ||||
Revenue rose 46.8% year over year (2023 ₩32.8 trillion → 2024 ₩66.2 trillion → 2025 ₩97.1 trillion), and the three-year trend is 'rising'. That said, the pace of growth slowed from the prior year. Operating profit rose 101.2% year over year. Over the 5 years on record, revenue compound annual growth (CAGR) is 22.6%. The two-year revenue CAGR is 72.2%. In the most recent quarter (Q1 2026), revenue was 198.1% higher than the same period a year earlier.
Latest quarterly results Source Q1 2026 · vs year-ago
Technical indicators Computed
What stands out
- P/E and P/B are both low versus peers, so the price looks inexpensive relative to earnings and assets.
- ROE of 45.8% points to solid profitability.
- Revenue grew 46.8% year over year, a sign of growth.
- The balance sheet is stable in terms of debt and liquidity.
Points to watch
- The figures shown are based on the last annual report as of the writing date, so it is best to review the latest quarterly results and filings alongside them.
Recent news & events searched · sourced
- 2026-10-30UpdateKioxia FY2026 Q2 results: October-December revenue and profit guided to roughly the same level as the prior quarter (currency effects amid strong data-center demand)A sign that a NAND maker's profit growth will pause next quarter (the company cited currency effects), and a variable for Q4 valuation gains or losses on the Kioxia-linked investment asset that SK Hynix holds through BCPE Pangea Cayman2. Source
- 2026-10-29UpdateClarifies reports of a new Solidigm NAND plant in the U.S. as 'under review, undecided'; re-disclosure due 2026-11-27The company said it is reviewing ways to expand capacity but nothing has been decided. If confirmed, capital spending would rise, but the company would gain a local production option ahead of U.S. chip tariff discussions. Source
- 2026-10-29UpdateSamsung Electronics Q3 results: sharp increase in HBM4 shipments, plans to expand HBM, server DRAM and SSD supply in Q4, and an outlook for 2027 AI demand to far exceed industry supplyA sign of stiffer competition for share among HBM suppliers, though the outlook for an industry supply shortage points the same way as SK Hynix's. Source
- 2026-10-28EarningsQ3 2026 preliminary consolidated results: revenue ₩104.6155 trillion, operating profit ₩83.3338 trillion, net profit ₩39.7756 trillionQuarterly revenue topped ₩100 trillion for the first time, and the 79.7% operating margin is a quarterly record. Net profit, however, fell 57.7% from the prior quarter because of −₩32.6253 trillion in non-operating items, opening a wide gap with operating profit. Source
- 2026-10-28IRQ3 earnings materials: non-operating items of −₩32.63 trillion (foreign-exchange-related losses ₩11.63 trillion, investment-asset valuation losses ₩22.92 trillion), quarter-end net cash ₩98.7 trillionLosses running opposite to the investment-asset valuation gains that lifted net profit in the first half were booked in Q3, showing that net profit can swing widely regardless of operating results. Quarter-end cash and equivalents were ₩114.46 trillion and borrowings ₩15.80 trillion. Source
- 2026-10-28IRQ4 shipment plan and long-term supply agreements: DRAM shipments to rise by about the mid-20s percent and NAND by the mid-single digits, long-term agreements with 14 customers covering more than 50% of 2027 and 2028 capacity, and demand expected to exceed supply through 2030Q4 DRAM volumes are set to rise sharply on the 1c process expansion and higher M15X output. Because the long-term agreements reflect the market within agreed price ranges, they act as a cushion when prices turn down and as a cap when prices spike. Source
- 2026-10-28FilingExpected fiscal 2026 shareholder returns of about ₩100 trillion: at least ₩30 trillion in cash dividends and at least ₩60 trillion in treasury share acquisitions and cancellations (including ₩40 trillion from August)The additional returns flagged in August have taken concrete shape. The company noted that the figure is forward-looking information and actual results may vary with business conditions. Treasury shares for employee compensation are excluded from the total, and the plan is to be carried out in stages through the payment of the year-end dividend. Source
- 2026-10-28FilingQ3 dividend of ₩20,000 per share (₩10,000 reflecting earnings + ₩10,000 Kioxia special dividend), ₩14.1035 trillion in total, record date 2026-11-30A large increase from the Q2 dividend (₩375 per share). Half is a special dividend distributing realized gains on the Kioxia investment, so it will not repeat every quarter; the rest of the special dividend is to be paid with the year-end dividend. Source
- 2026-10-28FilingDecision to acquire ₩20.0077 trillion in treasury shares (14,920,000 shares planned, 2026-10-29 to 2027-01-28)About ₩10 trillion is to be cancelled to enhance shareholder value and about ₩10 trillion used for employee share-based compensation. The company said the board will set the acquisition amount each quarter. Source
- 2026-10-26FilingCompletes the ₩40 trillion treasury share acquisition decided in August: 23,693,272 shares at an average of ₩1,688,426, cancellation scheduled for 2026-11-05Because the average acquisition price was above the reference price at the time of the decision (the August 18 close of ₩1,662,000), fewer shares were acquired than the roughly 24.07 million originally planned. Once cancelled, shares issued fall from 730,492,365 to 706,799,093 (share capital unchanged). Source
- 2026-10-23FilingConfirms the halt of Solidigm's pre-IPO fundraising and IPO reviewA definitive filing that follows the undecided clarifications (August 5 and September 4) on an August 5 report; the company said it took capital market conditions, among other factors, into account. The option of bringing in outside capital to fund Solidigm investment is off the table for now. Source
- 2026-10-21UpdateMeta Q3 results: low end of its 2026 capital spending range raised to $135-145 billion, with aggressive investment to continue in 2027Evidence that cloud investment, the direct source of demand for HBM and server DRAM in AI servers, did not slow in Q4. Source
- 2026-10-06FilingChange to Q4 transactions with SK ecoplant: expected procurement of ₩4.2462 trillion (₩4.0049 trillion for new fab construction)More than 20% above the amount disclosed on February 27. The figure includes both increases from new and changed projects and construction advances from changing supplier payment terms (from promissory notes to cash), so it cannot be read purely as a measure of fab construction speed. Source
- 2026-09-30UpdateMicron FY2026 Q4 results and outlook: 2027 and 2028 supply and demand tighter than in 2026, industry DRAM bit shipments growing in the low 20s percent in 2027-2028, and first wafer output from the first Idaho fab in mid-2027Competitors' new volumes increase meaningfully only after the second half of 2027, so until then industry supply growth is limited mainly to process migrations. Source
- 2026-09-18UpdateReplies to an inquiry on reports of a production base in Japan (undecided, re-disclosure 2026-12-17) and to reports of divesting a stake in a back-end facility in Chongqing, China (undecided, September 9, re-disclosure 2026-12-08)The company only said it is reviewing additional production sites and a restructuring of its packaging business; nothing has been confirmed. The direction of both will be set in the year-end re-disclosures. Source
- 2026-08-14UpdateHalf-year filing: conditions imposed by China's competition authority on the Intel NAND acquisition (five years from December 2021) are due to expire in December 2026Once the obligations to maintain pricing policy, expand production and support the entry of a third competitor into China's enterprise SSD market end, the company can apply for an exemption, and the authority will decide whether to grant it. It is a variable for how much room Solidigm has to run its China business. Source
- 2026-08-07InvestmentNew facility investment decisions: second fab at the Yongin semiconductor cluster, ₩35.2246 trillion (through 2031-10-31), and Cheongju M17, ₩19.1 trillion (through 2031-04-30)Equal to 29.19% and 15.83% of end-2025 shareholders' equity, respectively. They lay the groundwork for capacity growth but also raise depreciation and cash needs. Source
- 2026-07-15FilingResults of the U.S. American depositary receipt (ADR) issuance: 17,790,000 new shares, actual proceeds of ₩39.8905 trillion, all for capital spendingThe final issue price was ₩2,249,751, with payment on July 14. Shares issued rose to 730,492,365, and funding for large-scale expansion was secured in advance. Source
- 2026-06-29FilingFuture business and management plan (fair disclosure): medium- to long-term investment guideline of about ₩1,100 trillion (Yongin about ₩600 trillion, Cheongju about ₩100 trillion, southwestern region about ₩400 trillion)Individual investments will be disclosed separately when the board approves them. It is a long-term plan whose amounts and timing may change with market conditions. Source
- 2026-06-26UpdateDisclosure of a serious accident: one worker, on medical leave after an occupational illness claim was approved (March 2025), died (2026-06-25)The Ministry of Employment and Labor plans an on-site investigation and cause analysis, and the company will file a correction once it is confirmed whether the case qualifies as a serious accident under the Occupational Safety and Health Act. Depending on the conclusion, action by the authorities could follow. Source
- 2026-05-31UpdateNvidia announces full mass production of Vera Rubin, with production shipments from fall 2026A catalyst for HBM4 demand to take off in earnest. SK Hynix expects HBM4 to surpass HBM3E in both shipments and revenue in Q4. Source
- 2025-12-31UpdateThe U.S. Commerce Department's Bureau of Industry and Security (BIS) revocation of blanket authorization (VEU) for the China plants takes effect; 2026 operations run on annual equipment licenses (per the prospectus)The 2027 license must be approved again by year-end, and delays could disrupt equipment maintenance and replacement at the Wuxi DRAM and Dalian NAND plants. Source
Figure cross-check computed ↔ external
| Metric | Computed | External | Status | Source |
|---|---|---|---|---|
| Latest disclosure sweep (as of today) | fetch_disclosures days=200 2026-10-29, base major 15 | base 2026-06-09 → (2026-11-07) 100(2026-05-06~10-29). major 15: 10-28 ·10-28 ·10-26 ·08-21 ·08-19 (₩40 trillion)·08-07 ·08-07 2·07-29 2 ·07-22 ·07-15 ·07-10·07-06 ·06-29 ··06-24 . | Confirmed | link |
| Q3 2026 results checked against the original filing | DART : revenue 104 ₩615.5 billion·operating profit 83 ₩333.8 billion· 50 ₩708.5 billion·net profit 39 ₩775.6 billion· 39 ₩534.4 billion, 1~3 revenue 236 ₩510.5 billion·operating profit 181 ₩486.7 billion·net profit 173 ₩684.8 billion | 3 IR ·: revenue 104,615·operating profit 83,334· 50,708· 10,933·net profit 39,776, revenue 14,963· 6,319·D&A 4,341, −32,625 | Confirmed | link |
| Quarter-stacked forward estimate (2026 forward P/E) | Q1 37.61+Q2 60.54+Q3 83.33+Q4 101.0= operating profit ₩282.5 trillion → 4 ₩92.9 trillion· 22% → net profit ₩245.9 trillion → ₩1,021.23 trillion ÷ ₩245.9 trillion = 4.2x | — | Unverified | link |
| Updated denominators (P/E, P/B, ROE, debt ratio, margins) | TTM(2025Q4~2026Q3)·3 : PER 20.82x·PBR 5.58x·ROE 61.1%· 35.6%·operating margin 74.5%·net margin 70.1% | base: PER 23.21x·PBR 6.22x·ROE 45.8%· 35.6%·operating margin 58.6%·net margin 56.9% | Mismatch | link |
| Precise recalculation of the sub-sector key metric (P/B) | 1,021 ₩228.3 billion ÷ 3 309 ₩88.0 billion = 3.30x, 3.19x | base sector_valuation PBR 6.22x | Mismatch | link |
| Cash capacity (net cash, FCF yield) | 114 ₩460.0 billion − 15 ₩800.0 billion = 98 ₩660.0 billion; 4 FCF approx. 123 ₩500.0 billion ÷ = approx. 12.1% | IR p7· p16, 2025 ·3 | Confirmed | link |
| Enterprise value (EV) recalculation | EV = 1,021.23 − 98.66 = approx. ₩922.57 trillion; EV/EBIT 4.60x·EV/Sales 3.43x·EV/EBITDA 4.26x, operating profit EV/EBIT approx. 3.3x | base enterprise: −12 ₩413.8 billion·EV 1,008 ₩814.5 billion·EV/EBIT 13.04x·EV/Sales 7.64x·EV/EBITDA 11.04x·FCF 2.43% | Mismatch | link |
| Q3 non-operating items and the Kioxia-linked investment asset | 3 − operating profit = −32 ₩625.3 billion | BCPE Pangea Cayman2 Limited 66 ₩77.2 billion. 63 ₩179.0 billion· 13 ₩979.6 billion | Unverified | link |
| Consistency check of peer assumptions (memory) | SK 4 : D +25%· +8%, +5%· +3%, / −5% → operating profit +21% | 3 operating profit 107 ₩440.5 billion, 4 ' · ', HBM4· D·SSD . 4 net profit TTM PER approx. 6.0x·SK approx. 5.4x | Unverified | link |
| Company's Q4 plan (shipments, products) | D approx. 20% | M15X 2026 , 4 HBM HBM4 ·revenue HBM3E , 2027 HBM ·HBM D , 321 4 50% | Confirmed | link |
| Shares issued, treasury shares and market cap basis | ₩916.0 trillion | ₩916.0 trillion | Unverified | link |
| Default audit flags and consistency of the displayed forward P/E | base audit passed | base valuation.forward_per 8.6 ↔ PER 3.56 ↔ 4.2 | Mismatch | — |
| Largest shareholder stake (checked against default governance data) | SK 146,100,000 ÷ 730,492,365 = 20.00% | base governance 20.07% | Mismatch | link |
| Re-check of figures in the prose against source documents (QA) | 6 ·· · DART , IR ·, base . PER 1,021 ₩228.3 billion ÷ net profit ₩245.9 trillion = 4.15 → 4.2x | 10: 2027 HBM ' ', approx. ₩100 trillion '' → ' ', PER 10.42x , 13 ₩983.6 billion, EV , '50% ', HBM ' ', , , CXMT | Confirmed | link |
Recent filings Source
- 2026-06-09OwnershipOfficers'/major-shareholders' holdings report
- 2026-06-09OwnershipOfficers'/major-shareholders' holdings report
- 2026-06-09OwnershipOfficers'/major-shareholders' holdings report
- 2026-06-09OwnershipOfficers'/major-shareholders' holdings report
- 2026-06-09OwnershipOfficers'/major-shareholders' holdings report
- 2026-06-09OwnershipOwnership-change filing
- 2026-05-29Corporate governance report
- 2026-05-29Large-business-group status disclosure
- 2026-05-29Amended filing
- 2026-05-15OwnershipOfficers'/major-shareholders' holdings report
- 2026-05-15PeriodicQuarterly report
- 2026-05-13TreasuryMaterial-fact report
📖 Plain-language glossary — expand if you are new to this
- P/E
- How many times a year's net profit the price is worth (lower is cheaper relative to earnings). The P/E here is on trailing (last full-year) results; for companies whose earnings swing fast (memory chips and other cyclicals/high-growth), a forward P/E on this year's expected earnings is more accurate.
- P/B
- Price relative to net assets (equity). Around 1x means it trades near book value; below 1x means below book.
- P/S
- Price relative to a year's revenue — useful for growth companies with thin earnings.
- Net debt / EV
- Net debt = interest-bearing debt − cash. Negative means more cash than debt (net cash). EV (enterprise value) = market cap + net debt, closer to what it would cost to buy the whole business.
- EV/EBIT · EV/EBITDA · EV/Sales
- Enterprise value against operating profit (EBIT), EBITDA, or revenue. Unlike P/E these reflect debt and cash; lower is cheaper relative to earnings power or sales.
- FCF / FCF yield
- Free cash flow = operating cash − capex, the cash actually left over. FCF yield = FCF ÷ market cap; higher means more cash generated per unit of market value.
- Intrinsic value (DCF)
- Future free cash flow (or, for some capex-heavy but profitable names, forecast earnings) discounted to today to estimate per-share value. Because it shifts a lot with the discount-rate and growth assumptions, it is shown as a bear/base/bull range, and the basis and assumptions are disclosed in one line beneath it.
- ROE
- How much profit the company earns in a year on its equity (%). Higher means better returns on capital.
- EPS / BPS
- Earnings per share / net assets (book value) per share.
- Operating / net margin
- Profit left from the core business / final profit after tax and interest, per unit of revenue.
- Debt ratio
- Debt relative to equity (%). Higher means more reliance on borrowing (norms vary by sector).
- Current ratio
- Assets convertible to cash within a year against debt due within a year. Above 100% leaves some short-term headroom.
- Interest coverage
- How many times operating profit covers the interest owed. Below 1x means operating profit alone struggles to cover interest.
- Dividend yield / payout ratio
- The year's dividend as a % of today's price / the share of earnings paid out as dividends.
- Revenue CAGR
- Multi-year growth expressed as a single yearly average (compound annual growth rate).
- RSI (short-term signal)
- Whether recent price action is overheated or beaten down. Above 70 is overbought, below 30 oversold.
- MA20 / MA60 (moving averages)
- The 20- and 60-day average price. Price above them signals a firmer short-term trend.
- vs 52-week high
- How far below the past year's peak the price sits now (%).
All figures are for reference only; how they read varies by sector and over time.
Sources: Korea FSC market-price API (data.go.kr), OpenDART, KRX/KIND — public data only.
Bong Stocks presents public-data-based information for reference only. It is not investment advice and contains no target prices, ratings, or buy/sell recommendations. Verify independently before making any decision.