SK Hynix Q2 Earnings Cut but Target Prices Held: Synthesizing the Mirae Asset and KIS Reports

Context This post is a case where the frame discussed in Are Samsung Electronics and SK Hynix Really Oversold on a 2027 Consensus Basis?, that “the market has already priced in the worst-case scenario,” is actually put to the test by this earnings-cut event. It pairs well with HBM 2030 Supply-Demand Deep Research, Samsung Electronics and SK Hynix 2028E Profit Valuation, and Late-July Big Tech Earnings Calls and the Memory Thesis. The related hubs are the AI HBM Hub and the Exclusive Analysis Hub.

TL;DR

• Mirae Asset and Korea Investment & Securities (KIS) both cut their SK Hynix Q2 operating-profit estimates. Mirae Asset moved from KRW 70.7tn to KRW 62.3tn, while KIS came in at KRW 60.4tn. The reasonable range across the two reports is KRW 60-62tn, with a median of roughly KRW 61.4tn.
• The core story is not demand slowing down. It is a normalization: LTAs (long-term agreements) and a high HBM revenue mix mean the surging spot price is not fully reflected in blended ASP. Shipment forecasts are barely changed.
• That said, the cut is not confined to Q2. 2026-2027 profit and FCF were lowered as well. So it is more accurate to read this as the market rebasing profit expectations and a target P/B that had run too high, rather than as damage to the industry itself.
• Yet both brokers held their target prices (Mirae Asset at KRW 4.2m, KIS at KRW 3.8m). The reason is that the target price is derived from 12-month-forward BPS and a target P/B, not Q2 earnings. A valuation technique offset the earnings cut.
• The market does not yet believe that target price. On July 13, SK Hynix plunged -15.37%, and the gap is corroborated by upside to target of 106-128%. The rebound on the morning of July 14 is an oversold bounce, not a trend reversal. [Analysis scope]

1. What Changed: The Numbers From the Two Reports

1.1 Mirae Asset (Kim Young-gun, July 14, A Slightly Lower Gaze)

Maintains a Buy rating and a target price of KRW 4.2m, but lowers its earnings outlook. [Fact: Mirae Asset report]

The important point is that the DRAM shipment forecast is barely changed, and only the NAND shipment forecast was trimmed slightly. This is not a volume problem; it is a question of the timing and intensity of price recognition.

Mirae Asset’s supporting arguments are as follows.

• About 50% of revenue is estimated to run through LTAs, so the spot-price surge cannot be reflected immediately in full, but earnings stability rises in exchange.
• TSMC’s June revenue rose +67.9% YoY, and the combined backlog of five big-tech and neocloud companies stands at USD 2.1tn, indicating AI demand remains solid.
• Rising HBM prices and HBM’s encroachment on production capacity are also keeping commodity DRAM supply tight.
• 2027 operating profit still grows +45.7%, and 2028 is put at KRW 400.1tn, assuming no profit cliff.

In other words, this report lowers the immediate profit surge from the spot-price spike while keeping the long-run persistence of high earnings intact. The headline reads as a cut, but the substantive message is closer to a buy-the-dip call.

1.2 Korea Investment & Securities (Chae Min-sook and Kim Yeon-jun, July 13, Normalizing Estimates to Reflect LTAs)

Breaking down KIS’s estimate revisions, Q2 DRAM revenue and operating profit were cut -11.2% and -13.8% respectively, while NAND was actually raised +23.0% and +18.4%. This is not across-the-board memory demand weakness; it is an HBM/DRAM price-mix adjustment.

2. The Difference Between the Two Reports Is Time Horizon and Framing

Both houses cut estimates in parallel, but they emphasize different things. Understanding this difference also helps explain the market’s reaction.

Time horizon: Mirae Asset sees 2026-2027 profit higher than KIS does, but expects growth to plateau in 2028 (KRW 400.1tn). KIS is more conservative on the immediate Q2 and 2026 numbers, but sees profit growth continuing through 2028 (KRW 447.0tn).

Framing: KIS structures why it cut around LTAs and HBM mix, a fundamental reinterpretation. The logic is that a high HBM weighting means less of the spot-price surge carries into blended ASP, and that non-HBM DRAM assumptions were lowered. Mirae Asset, by contrast, argues that the share price fell more than the earnings cut warrants, a flow-reversal and buy-the-dip logic. Its view is that the stock over-corrected as earnings expectations and ADR-listing flow expectations unwound.

In short, Mirae Asset is more bullish across Q2, 2027, and the target price. Still, the two reports arrive at the same conclusion: cut near-term earnings while keeping the long-run thesis intact, with a Buy rating.

3. Earnings Were Cut, So Why Did the Target Price Hold?

This is the central question of this post. The answer lies in how the target price is derived.

3.1 The Mechanism: The Target Price Is Not Q2 Earnings

Target price = 12-month-forward BPS × target P/B. It is not based on Q2 earnings. KIS’s formula is fairly explicit. [Fact: KIS report]

12MF BPS KRW 643,124 × target P/B 6x = KRW 3.859m → target price KRW 3.8m

Three factors defend the target price here.

1. BPS roll-forward: as time passes, the 12-month-forward BPS anchor rolls onto a higher future book value, offsetting the near-term earnings cut.
2. The nature of the cut: this cut is a timing issue in non-HBM DRAM ASP, not damage to the duration of HBM earnings. The asymmetry shows this: on KIS’s numbers, 2027 DRAM operating profit was cut -16.9% while NAND was raised +12.8%.
3. LTA-based stability: the 6x multiple is applied on the premise that LTAs lower earnings volatility and that an operating margin in the 70%s persists through 2027-2028.

3.2 What Needs an Honest Look: What a 6x Target P/B Means

Holding the target price does not signal bullish conviction. A 6x target P/B exceeds the top of the historical band (roughly 5x), and it presumes a structural re-rating. In other words, “holding the target price” means the lower earnings were offset by the persistence of LTAs and a higher target P/B, not that earnings weren’t actually cut.

Mirae Asset deserves particular scrutiny here. The quantitative output shown in the report’s own table is as follows.

12MF BPS KRW 583,213 × target P/B 6.5x → report-derived target price approx. KRW 3.77m

Yet the official target price held at the existing KRW 4.2m. On the stated BPS, KRW 4.2m requires roughly a 7.2x P/B. In other words, Mirae Asset’s KRW 4.2m does not fully tie back to the report’s own quantitative formula. [Inference: back-solved from report table]

Applied strictly, this is how it breaks down.

The two brokers’ quantitative formulas converge to nearly the same place, KRW 3.77-3.8m. Only Mirae Asset’s KRW 4.2m departs from the quantitative basis.

4. The Market Does Not Believe This Target Price

4.1 The July 13 -15.37% Plunge and Flows

On July 13, SK Hynix plunged -15.37%, from KRW 2.18m to KRW 1.845m (Samsung Electronics fell -11%). That day’s flows were as follows. [Fact: local Kiwoom data]

Within institutions, investment trusts sold about -KRW 0.95tn and private funds about -KRW 0.27tn. This looks less like a reaction to a single report and more like a high-beta HBM pure-play de-rating in which foreigners, institutions, and program trading all cut positions at the same time. In fact, the selling had already started on July 10 (foreigners -775,000 shares, program trading -KRW 1.76tn).

4.2 The Wide Gap Between Target Price and Share Price

The crux of the market’s interpretation is the gap between the target price and the actual share price.

• Target prices of KRW 3.77m to KRW 4.2m imply roughly 106-128% upside from the current price of KRW 1.84m
• The current price implies roughly 3.0-3.3x P/B on the two brokers’ 12-month-forward BPS

Brokers are pricing a 6x P/B re-rating and earnings durability into their target prices, but the market does not accept that and trades around 3x. The conclusion reached by working backward in The Worst-Case Scenario Is Already Priced In, that “the market-implied EPS runs at roughly half of consensus,” is effectively borne out by this event. Brokers are pricing in an optimistic scenario, while the market is pricing in a 2028 normalization scenario.

4.3 The July 14 Rebound Is Not a Bottom Confirmation

On the morning of July 14, SK Hynix rebounded from a low in the KRW 1.75m range to around KRW 1.9m (roughly +3%). Samsung Electronics was stronger, up in the +5% range. [Fact: live quote] That is stronger than the KOSPI (around +1.7%) but weaker than Samsung Electronics, and still roughly -12 to -13% below the July 10 close.

This rebound is an overlap of Mirae Asset’s buy-the-dip call and an oversold dead-cat bounce, not a trend reversal. It is more accurate to read this as the first technical rebound now that the expiry and ADR-listing-expectation events are behind it.

5. Overall Assessment

The market’s interpretation can be summarized in four points.

1. This is not a memory-demand collapse. Strong bargain-hunting flowed in at the lows, and the shipment, AI-demand, HBM4, and LTA structures remain intact.
2. Prior earnings expectations had run too high. The 9-13% cut to 2026-2027 estimates is a real recalibration of expectations.
3. The market has not yet accepted a 6-7x P/B re-rating. The current price sits around 3x.
4. Some capital is rotating toward the Samsung Electronics catch-up trade. Hynix’s HBM purity is an advantage, but it has also become the first position to be cut whenever positioning gets overcrowded.

So this cut is not simple noise; it is a genuine recalibration of earnings expectations. But because the shipment, AI-demand, HBM4, and LTA structures are not impaired, it is not a confirmed peak-out signal either.

Between the two reports, Korea Investment & Securities’ KRW 3.8m target ties more cleanly to its formula and long-run estimates. Mirae Asset’s KRW 4.2m is directionally understandable, but based on the numbers actually presented in its own report, roughly KRW 3.77m is the more honest quantitative target. Today’s rebound is only a first signal of a possible bottom. Only if foreign and program selling eases into the close and the price recovers to around KRW 1.93m can we conclude that the market has begun to weight LTA-based earnings durability more heavily than the earnings cut.

6. What to Watch

Rather than simply following broker target prices, it is more accurate to track the market’s shift in interpretation directly through the following indicators.

• Flows: whether foreign and program net selling eases, or turns into net buying
• Price: whether the stock recovers to around KRW 1.93m (near the July 10 close)
• Fundamentals: at SK Hynix’s confirmed Q2 results in late July, the HBM sold-out period, the scope of 2027-2028 contracts, and commentary on HBM4E yields
• Macro: the direction of big tech’s 2027 capex guidance around July 28-30
• Relative: whether capital that rotated into the Samsung Electronics catch-up trade flows back

If these indicators align bullishly, it means the market is starting to accept the brokers’ earnings-durability scenario; if they align the other way, the target P/B re-rating itself is retreating.

This post synthesizes publicly available reports from Mirae Asset Securities and Korea Investment & Securities together with public quote and flow data to compare and verify the two reports. The target prices and investment opinions cited are the views of each brokerage; this article does not simply follow them, but critically reviews their consistency with the underlying quantitative formulas. The stocks mentioned are examples for analysis, not a recommendation to buy or sell any specific stock. Earnings estimates and target prices can change depending on when they were published, and Mirae Asset’s detailed valuation multiple is a figure back-solved from the report’s own table. Investment decisions and their outcomes are the sole responsibility of the investor.

Related Posts

• Are Samsung Electronics and SK Hynix Really Oversold on a 2027 Consensus Basis?
• HBM 2030 Supply-Demand Deep Research: Dissecting the 26.7EB Demand Model and Cross-Checking It Against the Capacity Expansion Schedule
• Samsung Electronics and SK Hynix 2028E Profit Valuation: Cheap-Looking Numbers and a Cycle Check
• Late-July Big Tech Earnings Calls and Memory Thesis Scenario Analysis
• Samsung Electronics’ Earnings Selloff and Its Nvidia-like Quarter