SK hynix (SKHY): Its PROFIT Was BIGGER Than Its REVENUE - Why We Say AVOID

SK hynix Inc. (SKHY) Q2 2026 — SK hynix (SKHY), the HBM leader that supplies the memory stacked next to nearly every AI accelerator, reported the largest quarter in Korean corporate history: revenue KRW 79.3T (+257% YoY), operating profit KRW 60.5T at a 76% margin (+557%), and net profit KRW 93.9T - a 118% net margin, meaning profit exceeded revenue. But it MISSED (street: KRW 84T revenue / KRW 64T operating profit), the margin beat came from commodity NAND (ASP +mid-50% QoQ) rather than HBM, and KRW 63.3T of that net profit is an unrealized fair-value mark on its Kioxia stake - a position with no board seat and no voting rights, on a stock that has since fallen 57%. Core net income ex-Kioxia is roughly KRW 45.5T. The Nasdaq ADR (listed July 10 in a record $26.5B offering at $149) trades near $127, about 31% ABOVE the identical Seoul shares, and the ordinary-to-ADR conversion quota was fully consumed at listing. Our mid-cycle model: $99 per ADS. Our call: AVOID. SK hynix just did something companies are not supposed to be able to do: it reported a quarterly profit LARGER than its quarterly revenue. Q2 2026 revenue of KRW 79.3 trillion (+257% YoY), operating profit of KRW 60.5 trillion at a 76% operating margin (+557% YoY), and net profit of KRW 93.9 trillion - a 118% net margin, and the fifth consecutive record quarter. The tax bill alone, KRW 28.8 trillion, was the largest single quarter of corporate tax ever paid by a Korean company. And yet three things are wrong with the picture. First, it was a MISS: the Street modelled KRW 84 trillion of revenue and KRW 64 trillion of operating profit, and SK hynix came in ~5.5% and ~6.6% light, because 'shipments of some high value added products were pushed into the second half' - the HBM4 ramp slipped. Second, the record margin was disproportionately a NAND story: NAND ASPs rose in the mid-50% range versus ~30% for DRAM, and NAND climbed from 21% to 27% of revenue while DRAM fell from 78% to 73%. Third, and biggest: KRW 63.3 trillion of that net profit is a Level-3 fair-value mark on SK hynix's Kioxia stake - held through a Cayman holding company with no board seat and no significant influence, roughly 80% unrealized - and Kioxia has since fallen ~57% from the June 30 price that set the mark. Strip it and tax it at the company's own 23.5% effective rate and core net income is about KRW 45.5 trillion, core EPS ~KRW 64,000 against the reported KRW 132,126. Roughly half the headline EPS is a Japanese share price. Then there is the wrapper. The Nasdaq ADR listed on July 10 in a $26.5 billion offering - the largest depositary-share deal in history - priced at $149, opened at $170, peaked at $194, and now trades near $127, about 15% below the IPO price. It also trades roughly 31% ABOVE the identical Seoul-listed shares, and because the ordinary-to-ADR conversion quota (17,790,000 shares, 2.50% of the company) was fully consumed at listing, that premium can only close by the ADR falling. Our valuation uses a mid-cycle earnings frame rather than a DCF, because capex in the high KRW 40 trillion range against ~KRW 16 trillion of annual depreciation makes near-term free cash flow meaningless: KRW 225T of mid-cycle revenue at a 40% through-cycle margin, taxed at 25%, at 13x plus KRW 95,200/share of net cash, gives KRW 1,299,000 per ordinary share - $99 per ADS after the 10:1 ratio, an FX rate of KRW 1,446.7/$, and a 10% structural ADR access premium. Today's Seoul price translates to about $97 per ADS, so two independent roads land two dollars apart, both ~22% below the Nasdaq price. Our call: AVOID, 2/5. This is not a bet against SK hynix the company - the franchise is an A, with 69.4 trillion won of net cash, ~56-62% HBM share and five-year agreements with roughly ten customers. It is a refusal to pay a 31% markup for a half-paper quarter at what may be the top of a memory cycle. We differ sharply from Barclays' Overweight and $330 target and from the Seoul consensus (11 brokers, Buy, KRW 4,060,000, ~$281/ADS) - their multiple is defensible, we just do not believe peak-cycle 2027 earnings are the right base. Interested near parity. Not financial advice. THE CALL: AVOID (2/5, A HISTORIC QUARTER, A HALF-PAPER PROFIT, AND A 31% MARKUP) — base-case value ~$99 vs ~$127 today. What to watch: 2027 HBM volume and pricing actually locked with the major customers and disclosed rather than alluded to, plus SK hynix raising the ADR conversion cap so the 31% premium can physically close - either would move us, both together and we would be buyers; the near-term risk to respect is the Kioxia reversal, since the KRW 63.3T gain was struck at a June 30 price and Kioxia has since fallen about 57%, which should swing Q3 non-operating income hard negative Also on YouTube: @ChargedAlpha DISCLAIMER: For informational and educational purposes only. Not financial advice. Do your own research before any investment decision.