Thornburg's Di Zhou Sees Memory Supercycle

SK Hynix’s US debut is off to a strong start, with ADRs indicated to open above the offering price. The portfolio manager cited a “memory supercycle” with strong fundamentals and SK Hynix as a top-quality pure-play memory name. Overall, sentiment appears supportive for near-term trading given the implied opening premium.

Analysis

The main signal is not the debut itself; it is that investors are willing to pay up for a Korean memory asset as if the HBM cycle has durability. That matters because it lowers the implied discount rate for the small set of memory leaders and can widen the valuation spread versus second-tier semiconductor names that still trade like cyclical commodity suppliers. In the next few weeks, this is more of a sentiment/flows event than a cash-flow event, but it can still push funds to add exposure to names with the cleanest leverage to DRAM pricing and AI server memory intensity.

Second-order winners are the equipment vendors and packaging/advanced substrate ecosystem that benefit when memory producers decide the cycle is real enough to keep spending. AMAT and LRCX should get the cleanest read-through if Hynix’s reception helps confirm capex discipline is breaking toward expansion; the lag is usually one to two quarters before revisions show up. The losers are downstream buyers with poor pricing power — consumer electronics, PCs, and some server OEMs — where memory cost inflation can quietly compress gross margin before it shows up in sell-side estimates.

The contrarian risk is that the market is extrapolating a structural supercycle from a single good tape. Memory rebounds often peak well before earnings do, because supply response eventually catches up and spot pricing rolls over faster than consensus expects. What would falsify the bullish read is flattening DRAM/HBM pricing, any cautious commentary from MU on inventory or customer digestion, or capex budgets that fail to re-accelerate over the next 1-2 quarters. For now, the cleanest view is to treat the IPO strength as a confirmation of positioning, not a reason to chase beta indiscriminately.

Key Decisions for Investors

• Buy MU on pullbacks over the next 1-3 weeks; this is the most direct U.S.-listed proxy for memory upside and should outperform SOXX if HBM pricing stays firm. Use a 10-15% relative outperformance target and cut if next earnings guidance implies inventory normalization.
• Add AMAT and/or LRCX on weakness as a 3-6 month capex continuation trade. The risk/reward is attractive if the market starts pricing a broader memory equipment upcycle; invalidate if memory pricing rolls over or capex commentary turns defensive.
• Pair trade: long MU / short WDC for the next 1-3 months to express DRAM/HBM strength versus more NAND-exposed memory exposure. This isolates the cleaner supercycle winner while reducing broad semi beta.
• Avoid chasing the ADR pop in the first 24-72 hours; treat it as a flow event. If the stock fades back toward offer price without a deterioration in pricing data, that is a better entry than momentum-buying.
• Set a watch item on DRAM and HBM spot pricing plus next major memory earnings guide. If pricing flattens or guidance turns cautious, take profits quickly — memory cycles can reverse faster than consensus expects.