Nanya Q2 2026 slides: pricing surge lifts margins to 79.5%

Nanya reported Q2 2026 net sales of NT$82.5B, up 68.2% sequentially and 684.2% YoY, driven mainly by pricing power (ASP +60%+ QoQ and +500%+ YoY). Margins surged with gross margin at 79.5% (from 67.9%) and operating margin at 73.7% (from 61.3%), alongside net income of NT$50.2B and EPS of NT$14.66. The company generated NT$55.0B of operating cash flow and NT$51.0B free cash flow after NT$4.0B capex, ending with net cash/equivalents of NT$198.4B post a NT$78.7B private placement. Management guided for further Q3 improvement and expects 2026 bit shipment growth in the high-teens YoY, underpinned by persistent AI-driven supply tightness despite major capacity ramp not contributing meaningfully until 2028.

Analysis

The real signal is not that memory is “strong,” but that the industry is behaving like a pricing oligopoly with capex still lagging demand. That supports a continued revision cycle for Micron (MU) and the broader semiconductor complex, but the equity upside is increasingly about earnings-duration expansion rather than immediate volume growth. The cleaner second-order winner is AI server infrastructure tied to higher-memory-content configurations; the cleaner loser is any OEM that cannot pass through rising DRAM content costs fast enough, especially PC and consumer-hardware names with low structural gross margins.

Near term, the market should treat this as a multiple-and-estimate setup, not a pure commodity trade. The next 1-3 months catalyst path is contract pricing commentary and Q3/Q4 guidance; if ASPs merely flatten while margins remain elevated, the trade can still work, but the beta will likely broaden from memory-only names into SOXX as earnings revisions spread. The falsifier is any evidence of inventory normalization or accelerated capex signaling from major DRAM vendors that implies the current shortage is being solved faster than expected.

Contrarian view: consensus may be underappreciating how much of the move is already discounted in the stock prices of the obvious memory winners, while underpricing the lagged margin pressure on downstream hardware assemblers. Over 6-18 months, the bigger risk is not geopolitics but self-inflicted oversupply: the announced multi-year capacity build-outs mean today’s scarcity can become 2028-2029 price competition if demand growth decelerates. That argues for preferring pairs and pullbacks over chasing the strongest tape.

Key Decisions for Investors

• Long MU on a 5-10% pullback; 3-6 month horizon; thesis is continued estimate revisions from pricing power, but size modestly because margins are already near peak-cycle optics.
• Pair trade: long MU / short DELL into the next earnings window; 1-3 month horizon; use the spread to express memory-cost inflation vs. OEM pass-through risk.
• Add a tactical long on SOXX only if the next round of semiconductor guidance confirms breadth beyond AI GPU names; otherwise avoid paying up for index beta on a single-factor earnings story.
• Do not chase BZWR or TGT on this print; there is no direct fundamental read-through from DRAM pricing to those names, so any move would be noise rather than signal.
• Set a falsifier alert: if contract DRAM pricing rolls over or inventory days rise in the next quarter, cut memory longs quickly; that would imply the current margin regime is closer to peak than the market expects.