Meet the Company Supplying 56% of the World's AI Memory Chips

SK Hynix entered the U.S. market with an ADS listing priced at $149 on July 10 (raising $26.5B), underscoring heavy U.S. appetite for AI memory. In Q1, revenue surged 198% YoY to 52.6 trillion KRW, with gross profit up 300% YoY to 41.7 trillion KRW and net income rising 398% to 40.3 trillion KRW, supported by HBM demand tied to AI compute. The main pushback is valuation: forward P/E edged to 10 vs Micron’s 6, implying shares may be pricey even as supply constraints keep prices elevated.

Analysis

The market takeaway is less about one company and more about how much AI memory scarcity is being capitalized into equities. Near term, U.S. retail and momentum flows can keep SKHY bid because the listing creates a fresh scarcity premium, but that is a sentiment-driven effect, not a proof of durable excess returns. The cleaner beneficiary on a 1-3 month view is NVDA: if HBM supply is genuinely tight, partner concentration reduces the risk of GPU shipment bottlenecks and supports data-center build schedules.

The bigger second-order issue is that fresh capital for capacity expansion can be self-defeating. If SKHY and peers spend aggressively into a still-hot cycle, the memory market can move from shortage to oversupply faster than investors expect, which is usually when multiples compress first and earnings follow later. MU is the read-through name to watch: it has less direct HBM brag value, but it is the most transparent proxy for whether the cycle is turning from scarcity rents to normalized pricing.

Contrarian view: the consensus is treating AI memory as a multi-year straight line, but memory historically re-prices on supply additions, not demand anecdotes. The first falsifier is any indication that HBM ASPs or lead times stop tightening over the next two earnings prints; the second is capex guidance that outpaces end-demand growth. For 6-18 months, Indiana/U.S. manufacturing diversification is structurally positive for political-risk discounting, but it also makes the supply response faster, which can cap the supercycle narrative sooner than bulls expect.

Key Decisions for Investors

• Do not chase SKHY after the listing pop; wait for a 10-15% pullback or evidence that HBM pricing is still tightening before initiating a long. Risk/reward is poor if the stock is already discounting a multi-quarter shortage.
• Relative-value trade: long MU / short SKHY for 3-6 months if SKHY continues to trade at a meaningful premium on U.S. sentiment rather than fundamentals. Thesis: the better-established U.S.-listed memory proxy should close part of the valuation gap if the AI cycle is real, while the ADR hype can fade.
• Long NVDA into any SKHY weakness as a supply-chain hedge: if HBM supply remains constrained, Nvidia's shipment cadence stays protected; if supply expands, AI deployment accelerates. Use a 1-3 month horizon and exit if NVDA commentary suggests memory is no longer a bottleneck.
• Watch list, not a trade yet: AMAT and LRCX on confirmation of SKHY capex spend. If the company follows through on manufacturing expansion, equipment names get a second-order order-flow tailwind, but the setup only works if fab spending is sustained.