Memory prices have gone up, iPhone orders are falling, and Samsung is cutting phone production by as much as 30%. Isn't that good for memory companies? Should I be worried if I hold chip stocks?
2026-10-12
A memory maker's profit is its customer's cost
Rising memory prices are profit for the sellers of memory and a cost for those who buy memory to build finished products. So far only the first effect has been priced into stocks. Two reports this week signal that the second effect is starting to show up in the numbers.
- Apple cut October component orders for the iPhone 18 Pro and Pro Max by about 15%. The two models launched at $1,199 and $1,299, $100 more than their predecessors, and demand for Apple devices fell from late August through October (FOX citing Reuters, 2026-10-09)
- Samsung's MX division asked suppliers to cut October-December volumes by 20-30%. Its annual production plan is expected to fall from about 270 million units to the low 200 millions (Money Today citing SamMobile, 2026-10-08)
- The cause is component costs. The price of 12GB LPDDR5X mobile memory was $145-146 in the second quarter, up 175% in a year, and Meritz Securities estimated that Samsung's MX and network divisions may have posted a 1.9 trillion won operating loss in the third quarter (SamMobile, 2026-10-08)
- TrendForce estimates memory's share of smartphone costs has risen from 10-15% to 30-40% (GSMArena, 2026)
When a component accounting for a third of costs nearly triples in price, device makers must either raise prices or give up margin. Apple raised prices and saw volumes fall; Samsung cut production to protect margins.
Who wins and who loses
| Category | Representative Stocks | Impact | Reason |
|---|---|---|---|
| Pure-play memory | SK Hynix, Micron | Near-term beneficiary, medium-term risk of slowing demand | Price gains are already in earnings. Smartphone and PC production cuts will come back as lower commodity DRAM volumes next year |
| Memory + devices | Samsung Electronics | Internal offset | MX losses partly eat into DS division profit gains. Third-quarter revenue coming in about 5% below expectations reflects the same structure |
| Devices | Apple, Xiaomi, PC makers | Hurt | Must choose between raising prices and losing volume |
| Device component suppliers | Camera modules, substrates, display components | Hurt the most | Customer production cuts translate directly into lower orders. No pricing power |
This is why Samsung Electronics cannot simply be seen as a memory beneficiary. Within the same company, one division's profit is another division's cost. SK Hynix and Micron, by contrast, have no device businesses and enjoy a pure benefit in the near term.
Is this a sign of a cycle peak?
History shows that customer behavior, not prices, is the first to flag a memory cycle peak.
- In 2018, cloud companies cut purchases after over-ordering in 2017 and building up inventory. NAND prices then fell about 60% and DRAM about 40%, and Micron shares dropped 57%, from about $64 in May to about $28 by year-end (Trefis, 2026-05-12)
- At the time, Goldman Sachs warned that "as price declines accelerate, customers delay purchases to wait for lower prices, a snowball effect that can last several quarters" (Motley Fool, 2018-09-11)
The differences this time are significant. In 2018, cloud companies, the source of demand, cut their own orders. Today it is smartphones and PCs that are cutting, while AI servers, the growth engine for memory demand, have not yet reduced orders. So the current signal is closer to "slowing growth in commodity memory" than "a peak". Still, if prices keep rising while demand slows, customers could cut inventory and delay purchases as in 2018. The indicator to watch is whether the rate of increase in fourth-quarter mobile DRAM fixed contract prices turns lower.
What should investors do?
- Semiconductor ETF holders: Memory is not a large weight in SMH, which is dominated by AI accelerators and foundries. Smartphone production cuts are a more direct risk to individual memory stocks than to the ETF as a whole
- Micron: In the signal history, its October 9 close of $1,029 fell below the 10-day moving average of $1,061.88. It remains above the 50-day ($976.19) and 150-day ($810.26) averages, so the entry condition (entry_ok) still holds, but it is -18.0% from its high of $1,254.81 with only 12.0% of room left before the trailing stop. New entries are better made after confirming a recovery of the 10-day average
- Samsung Electronics: Because the memory boom is partly offset by MX losses, check the MX fourth-quarter shipment guidance on the October 29 conference call
- Device component suppliers: Apple and Samsung production cuts will show up directly in fourth-quarter results. Reducing exposure until earnings are out is the safer course
Detailed charts and signals for other stocks are available on the /signals dashboard.