JPMorgan: Korean Memory Stocks Enter Expectation Reset Season, Pullback Creates Mid-Term Buying Opportunity
nashnova research
J.P. Morgan said on October 15 that Korean memory stocks have fallen roughly 15% over three months, diverging sharply from the Philadelphia Semiconductor Index; the bank sees this pullback as a mid-term buying window and recommends accumulating on weakness.
What are investors actually arguing about?
Analyst Jay Kwon visited over 100 Asian investors across Singapore, Kuala Lumpur, and Hong Kong in two weeks. He found one core divide.
Top-down investors worry AI capex may not last; bottom-up investors track memory supply-demand data and see fundamentals intact. This means → the same stocks produce opposite conclusions, and that sentiment split is itself a source of volatility.
In plain terms = one camp fears "AI hype fades, hyperscalers cut orders"; the other sees "chips are still in short supply." Earnings will decide who is right.
How far have memory stocks fallen — and versus the benchmark?
Korean memory stocks dropped roughly 6% in the past month and about 15% over three months.
Over the same periods the Philadelphia Semiconductor Index rose about 10% and fell about 4%, respectively — a stark divergence.
This means → memory has underperformed the broader chip sector, and the gap is widening. J.P. Morgan labels Q3 an "expectations-reset quarter" — the market is recalibrating earnings assumptions for memory. Near-term catalysts are scarce, but the bank argues this price dislocation gives mid-term bulls a "duration trade" entry.
In plain terms = a "duration trade" is a bet that the memory upcycle runs longer and higher than the market is pricing — not a play on tomorrow's bounce, but on a six-to-twelve-month trend.
Which names does J.P. Morgan favor?
Core overweights for the next six months: Samsung Electronics and SK Hynix (memory mainline), Kioxia (memory), Samsung Electro-Mechanics (MLCCs and substrates), ISU (AI printed circuit boards), and ISC (test sockets).
This reflects a systematic play across the memory supply chain — chips, packaging substrates, and test equipment — rather than a single-link bet.
In plain terms = the logic is: if the memory cycle really runs "longer and higher," component suppliers ride the wave alongside chipmakers.
What do the early earnings signals tell us?
Three sets of preliminary results are in: Samsung Electronics met expectations, LG Electronics missed, and LG Energy Solution beat.
This means → sector divergence is already visible. J.P. Morgan flags tech and financials as the key swing sectors this earnings season; life-insurance earnings delivery also deserves tracking.
The week of October 19 brings a cluster of full results — the first real test of the thesis.
Under what conditions does this call break down?
J.P. Morgan states its falsification conditions explicitly: if memory supply-demand keeps weakening, catalysts remain absent, or earnings see broad-based downgrades, the bullish call should be withdrawn.
Key tracking metrics include memory spot prices, AI-related order flow, and marginal changes in capital expenditure.
The verification window runs from October 19 through early November 2026 — beyond that window, the call's validity needs reassessment.
市场有风险，内容仅供研究参考，不构成投资建议。