Goldman Sachs sees a potential opportunity in Samsung Electronics and SK Hynix because it believes investors have overestimated how quickly Chinese chipmaker CXMT can catch up in advanced memory technology. That is a medium-term risk/reward view, not evidence that the shares have bottomed. Goldman’s assessment also distinguishes a credible capacity-expansion risk from an unproven near-term technology threat.
17
18
Why Goldman thinks the selloff may have gone too far
CXMT president Cao Kanyu said the company plans to introduce a server DDR5 registered DIMM using a 4F² memory-cell architecture by the end of 2026. The design aims to increase DRAM density, but the announcement did not include detailed product specifications or yield data.
2
3
That makes the announcement a potential competitive signal, not proof that CXMT can produce the chips at high volume, achieve reliable yields or win customer acceptance. The product target alone also does not establish that CXMT has matched Samsung and SK Hynix in advanced DRAM or high-bandwidth memory (HBM). Goldman’s reported view is that markets have overstated the speed of that near-term catch-up, while concerns about CXMT adding capacity are more credible.
17
18
In other words, Goldman’s case rests on the distinction between more manufacturing capacity and a demonstrated ability to compete across advanced memory products. The first can pressure the industry over time; the second is not established by a product announcement without supporting production and customer details.
The market is weighing more than CXMT’s technology
A report on Goldman’s call said major memory stocks—including Samsung, SK Hynix, Micron, SanDisk, Kioxia and CXMT—had fallen between 11% and 26% since October 10, erasing more than $700 billion in market value.
4 Those figures capture how broadly investors reassessed the sector, not just the two Korean companies’ direct exposure to CXMT.
The concerns include the possibility of future oversupply as well as competition from China. CNBC described a potential 2029 oversupply as one of the market’s worries, separate from the immediate reaction to CXMT’s announcement.
14 A longer-term supply risk can weigh on valuations even if the announced product does not immediately change the competitive landscape.
Goldman’s reported valuation argument is that prices already reflect considerable pessimism. The available reporting supports that broad thesis, but does not independently confirm the specific 2027 price-to-book multiples or the precise timing of a downcycle.
17
18 Those details should therefore not be treated as established here.
What supply and trading signals suggest
Rosenblatt analyst Kevin Cassidy said CXMT’s developments were unlikely to materially change memory supply and demand in the near term.
14 Separately, Roundhill Investments CEO Dave Maza described supply as tight, with chips sold out through 2027 and customer demand extending into 2028. That is an industry executive’s assessment, not a guarantee that shortages or high prices will persist.
31
The market also continued to sell after Goldman’s supportive view was reported: on Friday, U.S.-listed SK Hynix and SanDisk each fell about 3%, while Micron fell about 2%.
13 That shows Goldman’s thesis had not settled investor concerns about the wider sector or the timing of a recovery.
Is the memory-chip rout justified?
The evidence supports skepticism that CXMT’s announcement alone proves an imminent collapse in Samsung’s or SK Hynix’s competitive position. The product is still a target for late 2026, important production details were not disclosed, and Goldman argues the market is overestimating CXMT’s near-term technology catch-up.
2
3
17
18
But that does not make the whole selloff irrational. Capacity additions and the possibility of a later oversupply are distinct risks, and negative sector sentiment persisted even after Goldman’s assessment.
14
13 The most balanced reading is that the market may be pricing in too much immediate damage from CXMT, while still discounting real uncertainty about future supply, demand and the memory cycle. A more attractive risk/reward, in Goldman’s view, is not the same as a guaranteed rebound.