SK Hynix Fell 5% the Day JPMorgan Launched a $245 Bull Case — The Drop Wasn't the Signal, the 7x Multiple Is

SK Hynix (SKHY) fell 5% to $189.47 on Thursdayeven as JPMorgan initiated coverage with a $245 price target for June 2027. A stock sliding on a high-profile bullish call reads as a contradiction. It is not, once you separate the two things the headline fuses together: how far the price had already run, and what actually moved it that day.

What the 5% wasn't

Start with the price. The ADR was falling from close to $198.63, within a dollar of its $199.87 52-week high, after a 40% run in a single month. A 5% step back from the top of that move is a reversal inside an uptrend, not a repudiation of the bull case. The drop also was not Hynix-specific. The benchmark 10-year Treasury yield reached 4.91% while the 20- and 30-year yields sat at 5.3%; WTI crude pushed above $100 a barrel on renewed Middle East conflict; and markets braced for a rate hike at that week's Federal Reserve meeting. Foreign investors in Seoul werenet sellers.

That macro risk-off hit the memory complex as a group: Micron fell 3%, Western Digital fell 3%, and the Roundhill Memory ETF fell 3%, while the S&P 500slipped just 0.57%. That breadth is the signature of beta and momentum being unwound in the highest-multiple names, not of investors walking away from high-bandwidth memory. Korea transmits the move directly — SK Hynix and Samsung make up more than half the KOSPI, so a risk-off session in Seoul flows straight into the US-listed ADR.

The real number is 7x earnings

The durable figure on the tape is neither Thursday's 5% nor JPMorgan's $245; it is the valuation both sit on. JPMorgan built its target fromroughly seven times the average earnings it expects for fiscal 2026-27, and the ADR itself trades around 7x earnings. Two readings attach to a single-digit multiple for a company reporting ~76% gross margins and roughly 60% return on invested capital.

The bull reading, which is JPMorgan's: this is the dominant high-bandwidth memory supplier — it holds roughly half or more of the HBM market and counts Nvidia as its anchor customer — generating billions in free cash flow with a net-cash balance sheet, yet the stock trades at 7x earnings because the market reflexively distrusts memory earnings. JPMorgan argues the AI-driven memory upcycle lasts more than five years, memory average selling prices rise through late 2028, earnings per share compound at 34%, and more than half of SK Hynix's capacity is already locked under long-term agreements, deliveringan estimated 42% total shareholder return between 2026 and 2028.

The skeptical reading — what a single-digit multiple on record profits actually prices — is that this is peak-cycle earnings. Memory is the textbook boom-bust business: record profits fund capacity; the capacity floods the market; prices collapse; the cycle resets. JPMorgan's own $245 target is only about 24% above the recent all-time high and roughly 29% above Thursday's $189 close. When a stock that just ran 40% in a month reaches a bull target that sits barely 30% higher and prices the whole thing at 7x earnings, the initiation's real content is reassurance that the cycle is not ending — not a runway to a new valuation regime.

Two hands, one company

That is why the company's own capital decisions matter more than any single analyst note. SK Hynix did two things that point in opposite directions at once. In August it approved a roughly $28.6 billion buyback and cancellation covering about 3.3% of shares outstanding —the largest treasury cancellation in South Korean corporate history — and committed to returning more than half offree cash flow. In the same stretch it committed to at least$31 billion in capital spending this year, approved roughly$38 billion for two new memory fabs, and saw its CEO warn that 2027 could bring change. In memory, capex is the bear case in direct proportion to how strong the bull case is: every boom funds the oversupply that ends the boom.

So do not read Thursday's 5% as the market calling JPMorgan wrong. It was a macro-driven, sector-wide profit-taking event on an over-extended stock. But do not read the $245 stamp of approval as the story either. The story is that a company whose dominant HBM position is undisputed trades at 7x earnings — the market pricing peak-cycle profits while the bull side argues for a multi-year structural upcycle. The gap between those two is where both the risk and the reward sit.

The live question is therefore whether memory pricing holds as HBM4 ramps into 2027 for Nvidia — the forecast the entire bull case turns on. If average selling prices keep rising through 2028 as JPMorgan projects, 7x earnings on growing profits is cheap. If the industry builds into oversupply and prices roll over, the low multiple is exactly what it should be. Thursday's selling says the market is currently betting the second. That tension, not the analyst note, is the whole of the ADR today.