SK Hynix: Memory Production Sold Out Through 2029 — Bottleneck Could Last Until 2030
Samsung delivered its best quarter ever — but the market sold the stock anyway. On 29 October 2026, the company reported operating profit of 107.44 trillion won (roughly $80 billion, approximate conversion), the first technology company to exceed 100 trillion won in a single quarter. The day before, rival SK Hynix said DRAM and HBM production is effectively sold out through 2029. Yet Samsung shares fell as much as 6.2%, and the two companies themselves disagree on how long the shortage lasts: Samsung says the market will remain tight "through 2028," SK Hynix "through 2030." This disagreement — not the numbers — is the heart of the story.
The Numbers: A Quarter Built on One Division
Samsung Electronics reported consolidated revenue of 195.3 trillion won (approximately $145.8 billion) and operating profit of 107.44 trillion won (approximately $80 billion) for the quarter ending 30 September 2026 — up 783% from a year earlier, according to the company's official earnings as reported in secondary sources (aae691b8). Currency conversions vary slightly between sources and should be read as approximations. It is the fourth consecutive quarter of record profit and revenue, according to the Associated Press (142dfc04).
Nearly all of this came from semiconductors. The Device Solutions division accounted for 106.5 trillion won in operating profit — up from 12.2 trillion won in the same quarter last year (1067eedf). A note on the arithmetic: one headline describes this as a "15-fold jump," but the cited figures imply an eight-to-nine-fold increase (106.5 divided by 12.2 is roughly 8.7). It is still an enormous leap, but "15 times" is not supported by the underlying numbers.
The rest of Samsung lost money. The combined consumer electronics division — mobile phones, network equipment, TVs and home appliances — posted an operating loss of 800 billion won (approximately $597 million), of which the mobile and network segment accounts for around 600 billion won (aae691b8). In other words: Samsung is no longer, on its income statement, primarily a consumer brand company — it is a memory factory with a consumer business in the red around it.
SK Hynix the Same Week: Sold Out Through 2029
The day before Samsung's report, on 28 October, SK Hynix delivered its own record numbers: revenue of 104.62 trillion won (approximately $78 billion), up 328%, and operating profit of 83.33 trillion won (approximately $62 billion), up 632% (154653a0).
The company said DRAM and HBM volumes for the next three years, through 2029, are effectively sold out. Long-term agreements (LTAs) have been signed with 14 key customers, including major global cloud service providers: "We have to date signed LTAs with 14 key customers, including major global CSPs," the company said at its earnings call, according to PCMag. "While LTAs typically have a five-year term, we are actively discussing extensions of existing contract periods with some customers" (9edb383e).
CFO Woo-hyun Kim laid out the longest forecast in the industry: "Given the complexity of implementing leading-edge process nodes and the lead time required to build new production facilities, a situation where supply cannot keep up with demand is expected to persist for some time — a constraint expected to continue through 2030" (154653a0).
The Paradox: Why Did the Stock Fall?
Record earnings and sold-out capacity should normally lift a stock. Samsung shares fell as much as 6.2% after the report, and the company postponed its next shareholder return policy to January (1067eedf).
The closest explanation in the sources is doubt about duration: the market is paying for a level of profit that depends on a historically large memory shortage, and it has begun to price in when that shortage ends. Three factors point in the same direction:
- Samsung's own horizon is shorter than SK Hynix's. Samsung says the market "should remain tight through 2028" because new fabrication capacity takes time to build (1067eedf). That is two years shorter than SK Hynix's 2030 forecast. This is a genuine disagreement between the companies, not an error in the data — but the market may read Samsung's shorter horizon as a signal that the peak is nearer.
- Price momentum is flattening. TrendForce expects conventional DRAM contract prices to rise 10–15% in the fourth quarter, versus roughly 60% in the second quarter (1067eedf). Price growth is still strong, but the steepest phase may be behind us.
- The postponed shareholder return policy suggests even management is unwilling to commit to a return scheme in the middle of a record revenue run.
What the Companies Are Doing: Capital, Contracts and 2nm
Samsung increased chip capital expenditure by 62%, to 76.8 trillion won (1067eedf). That is a double signal: the investment is meant to capture AI demand, but at the same time increased supply is the mechanism that eventually dampens the very prices the company profits from.
On the contract side, Samsung is following SK Hynix's LTA model: 10 long-term agreements are signed with major customers, negotiations are under way with 10 more, and the goal is to have 60–70% of production capacity under such agreements by 2030 (1067eedf). At the same time, the company says major customers are asking for more deliveries even though next year's HBM production is already fully booked (1067eedf). Samsung expects AI demand to accelerate through 2027 and exceed supply (142dfc04) — notably not "through 2028," which is the horizon in the tight-market assessment reported separately.
The foundry business is a separate growth story. On the earnings call, Sukchae Kang, executive vice president at Samsung Foundry, said: "We have secured 2nm projects from major CSPs and AI HPC customers in Q3 as well, as we continue to expand contract wins and our design win portfolio on advanced process nodes. Notably, among advanced node projects, our 2nm agreements have increased significantly" (aae691b8, via TheFly transcript). That places Samsung in the fight to manufacture the AI chips themselves, not just the memory around them.
Open Questions: Duration, Concentration and Dampers
When does it turn? The difference between "tight through 2028" and "a constraint through 2030" is not a technical detail — it is two years of extraordinary margins. Both companies point to the same mechanism (lead times for new fabs, the complexity of leading-edge process nodes), but they draw different conclusions about how quickly new capacity catches up with demand. The sources provide no basis for determining who is right.
The concentration risk. Samsung and SK Hynix together control roughly two-thirds of the global memory market (142dfc04). When two companies stand behind almost the entire world's AI memory supply, the cost base of the entire global AI build-out becomes sensitive to their capacity decisions. South Korea's central bank (Bank of Korea) has, according to the AP, warned against the economy becoming too dependent on chips — a macro risk outside the companies' control (142dfc04).
What could actually dampen demand? The sources say customers are "racing to expand AI infrastructure" (142dfc04) and that memory is sold out with customers asking for more (1067eedf). They do not answer how long this race lasts, or what happens if the hyperscalers adjust their investment. TrendForce's price data points in the same direction as Samsung's shorter horizon: the steepest price growth appears to be flattening, but demand still exceeds supply.
The missing micro-detail. Capacity figures per node, the specific customer names behind the LTAs and detailed per-segment price development are not documented in the available source material. That limits the ability to model precisely when the shortage closes.
What Will Decide the Story
Three things to watch going forward, based on the sources:
- Price growth in Q4 and beyond — whether TrendForce's 10–15% holds, or rises back toward the Q2 level (indicating continued acute scarcity) or falls further (indicating normalization).
- The capacity effect of Samsung's 62% increase — when new fabrication capacity from 76.8 trillion won in capex actually comes online, it will test both Samsung's own 2028 horizon and SK Hynix's 2030 forecast.
- Whether Samsung reaches its 60–70% LTA target by 2030 — the more capacity locked into long-term agreements, the less spot-market volatility hits the company, but also the less optionality it retains if prices keep rising.
Samsung has delivered the largest quarterly profit any technology company has ever posted, and the market has responded by selling. That is less a paradox than a price signal: the stock market is not pricing the boom, it is pricing its end — and the two companies that know it best disagree on when that comes.