Samsung Electronics Co., Ltd. (KRX:005930) closed at ₩255,500 on Friday, September 4. It rose 2.2% on a day the KOSPI rose 1.64%. The company's market capitalisation was ₩1,493.7 trillion at Friday's close. That's roughly $1.07tn at an approximate ₩1,400 per dollar. That figure sits against ₩579.3tn of book equity at the end of June. That makes the stock worth about 2.6 times what its balance sheet says it owns.
Two and a half times book is not a number Samsung has often traded at. Ten years of price-to-book data from GuruFocus put the low at 0.9. The median is 1.41. The average is 1.6 and the high is 3.45. Today's multiple is closer to the top of that range than to the middle. On earnings, the same stock looks cheap. Trailing twelve-month net income is ₩150.7tn, which puts the shares near 9.9 times. Finbox puts the ten-year average price-to-earnings ratio at about 15.
So which is it? My view is that the book multiple is the honest one. Samsung's equity compounds through every cycle. Its earnings swing by a factor of three or four from trough to peak. When the two multiples disagree this widely, it's because the market is pricing the earnings as temporary. The rest of this piece is about whether the history supports that.
The eleven fiscal years on file tell a repeating story. Net income was ₩19.1tn in fiscal 2015. It climbed to ₩42.2tn in 2017 and ₩44.3tn in 2018, the last memory boom. It then halved to ₩21.7tn in 2019. It rose again to ₩55.7tn in 2022, the year memory prices peaked during the pandemic buildout. It fell to ₩15.5tn in 2023, the worst year of the decade. Memory prices collapsed that year.
Fiscal 2025 brought ₩45.2tn, back near the 2018 level. Then the first half of 2026 produced ₩118.8tn of net income in six months. That's more than double the best full year the company has ever recorded. Operating income in the second quarter alone was ₩89.5tn, against ₩4.7tn in the same quarter a year earlier.
Every prior peak in this series lasted about two years before earnings fell by half or more. That's what a memory cycle looks like from inside Samsung's income statement. The current one is much taller than the others. But it has the same shape so far: a steep rise from a deep trough.
The market has seen this movie. It is paying 9.9 times trailing earnings because it doesn't expect those earnings to persist. If it did, the multiple would be closer to 15, and the market cap would be closer to ₩2,260tn. The gap between those two numbers is the discount for cyclicality.
The reason to anchor on book rather than earnings is that book doesn't reverse.
Total equity was ₩179.1tn at the end of fiscal 2015. It reached ₩247.8tn by 2018 and ₩354.7tn by 2022. It reached ₩436.3tn by 2025 and ₩579.3tn at the end of June 2026. Through two memory busts, it never fell. Even 2023, with net income of ₩15.5tn, added ₩9tn to equity after dividends.
That's why price-to-book has been the more stable way to value Samsung across a decade. At 0.9 times book, the stock was in the worst part of the range. The market was saying the company would earn less than its cost of capital for a long time. At 3.45 times, it was pricing sustained returns far above it. Both extremes reversed.
At 2.6 times today, the market is pricing something well above normal but short of the decade's peak. Getting back to the median of 1.41 would require book equity to roughly double while the price stayed flat. Or the price would need to fall by nearly half. The first path takes several years of retained earnings at the current pace. The second takes one bad memory cycle.
There's a second reason to take the cyclical reading seriously. It's in the working capital lines.
Trade receivables were ₩96.4tn at the end of June, up from ₩51.1tn at the end of 2025. Inventories were ₩71.4tn, up from ₩52.6tn. Both lines have nearly doubled in six months. That's what happens when memory prices spike. The same physical volume of chips is worth far more on the books. Customers take longer to pay for it.
Cash and equivalents rose to ₩92.9tn from ₩57.9tn. Total assets reached ₩759.5tn from ₩566.9tn. The company has grown its balance sheet by a third in half a year. That came almost entirely through price, not volume. Capital spending on property and equipment was ₩31.2tn in the first half, up only modestly from the prior year.
I wrote on August 27 that more cash went into deposits than into factories. That still holds. A company that believed the current pricing was a new floor would be building capacity to meet it. Samsung is banking the proceeds instead. Its own capital allocation matches the market's earnings multiple, not its book multiple.
The case against the cyclical reading deserves a fair hearing. The current cycle is different in at least two ways.
First, the demand driver is AI infrastructure rather than smartphones or PCs. High-bandwidth memory sells under longer contracts to fewer customers. It has less spot-market pricing than commodity DRAM. If a larger share of Samsung's memory revenue moves onto that kind of contract, earnings could stay elevated longer than the 2018 or 2022 peaks. The half-year report doesn't break out HBM, so nobody outside the company knows the share.
Second, the competitive set has narrowed. Three companies make most of the world's DRAM, and capacity additions take years. Samsung's first-half figures suggest all three may keep capital spending restrained. If they do, supply may not catch up in the usual two-year window.
Against that, China's YMTC has stated a goal of becoming the largest NAND maker by end-2027, which I covered separately. And the Bank of Korea has raised rates twice in response to the chip boom. That's a reminder that a stronger won cuts into every won figure the company reports.
If the peak holds, 9.9 times earnings is the right multiple and 2.6 times book is cheap. If it doesn't, the book multiple is the one that has the further to fall.
The third-quarter preliminary earnings are due in early October. They will show whether second-quarter operating income of ₩89.5tn was a peak or a plateau. Samsung releases these as a fair disclosure (공정공시) about a week after quarter-end. It's a short filing with revenue and operating income only. A sequential increase keeps the plateau case alive. A sequential decline, even a small one, is what the 9.9 times multiple is already expecting.
Beyond that, watch the receivables line in the full quarterly report. If receivables keep rising faster than revenue, customers are stretching payment terms at the top of the price cycle. That's how the 2018 and 2022 peaks ended. If receivables flatten while revenue holds, the market's earnings multiple is too low.
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.