UBS Cuts SK Hynix Earnings by 19%, Lowers Target Price to KRW 3M but Maintains Buy Rating
0xBroomberg
UBS cut its 2027 operating-profit forecast for SK hynix by 19% to KRW 504.5 trillion and lowered the target price from KRW 3.2M to KRW 3M, yet maintained a Buy rating — even after the cut, UBS's 2027 estimate sits 17% above consensus, implying the market has already priced in a worse scenario.
Earnings slashed by a fifth — why still say "Buy"?
UBS cut SK hynix's 2027 operating profit forecast by 19% to KRW 504.5 trillion, and 2028 by 18% to KRW 544.5 trillion.
This means → even post-cut, UBS's 2027 number is still 17% above market consensus. The downgrade moves the call from "far above expectations" to "still above expectations."
Target price lowered from KRW 3.2M to KRW 3M; Buy rating unchanged.
Where did Q2 miss?
SK hynix reported Q2 2026 revenue of KRW 79.3 trillion, up 51% QoQ and 257% YoY; operating margin hit 76.3%, roughly in line with consensus.
But it fell short of UBS's own forecasts of KRW 86.5 trillion in revenue and KRW 68.5 trillion in operating profit. The gap came down to DRAM average selling price: the company disclosed a QoQ rise of roughly 30%, versus UBS's prior assumption of 42.6%.
In plain terms = volumes were fine, but the price per memory chip didn't rise as fast as UBS expected — a lower-than-modeled ASP dragged profits.
Why did the ASP fall short?
Mobile DRAM revenue rose to 19% of the DRAM mix. Mobile DRAM carries a lower ASP than server DRAM, pulling down the blended average.
Some large-customer long-term agreements (LTAs) — contracts that lock in price and volume — kicked in, preventing the company from fully riding the spot-price rally.
HBM4 — fourth-generation high-bandwidth memory, a premium product built for AI chips — began volume shipments but was concentrated in the back half of the quarter, limiting its mix uplift.
This means → UBS lowered its Q3 DRAM ASP growth forecast from 21.1% to 18.8% and Q4 from 12.6% to 8.4%, accepting that rising LTA coverage reduces near-term pricing elasticity.
How long can the shortage logic hold?
UBS projects 2027 industry-wide DRAM end-demand growth accelerating from 22% in 2026 to 36%, and NAND from 20% to 23%. On the demand side, AI agents are expected to drive DDR5, LPDDR5, and NAND capacity needs simultaneously.
The supply constraint is structural: HBM is projected to consume roughly 500K wafers/month of front-end DRAM capacity by end-2026, or 25% of the industry total; by end-2027, 690K wafers/month, or 31%.
In plain terms = the more HBM booms, the less new capacity is left for conventional DDR memory. That industry-wide squeeze — not just HBM itself — is the real floor under earnings growth.
Will SK hynix's HBM lead last?
UBS forecasts SK hynix's 2027 HBM bit-shipment share dropping from 48% to 39%, just below Samsung's 41%; Micron at 20%.
This reflects a framework of "market expands, the leader stays strong, but share normalizes" — not a permanent-monopoly model.
This means → the investment case doesn't rest on SK hynix owning HBM alone. It rests on total HBM market expansion plus a persistent squeeze on conventional DRAM supply displaced by HBM — a dual support for earnings.
With capex this high, is there still cash for shareholders?
UBS raised its capex forecasts: KRW 47.1 trillion in 2026, KRW 62.3 trillion in 2027, KRW 67.4 trillion in 2028. The first cleanroom at the Yongin fab is expected to take equipment in February 2027.
Even so, UBS projects free cash flow of KRW 187.5 trillion, 320 trillion, and 373.8 trillion for 2026–2028 respectively.
UBS speculates the company may launch a roughly KRW 12 trillion buyback in H2 2026 and could eventually return about 50% of FCF via dividends and buybacks — but neither the buyback size nor the 50% ratio is an official company commitment.
This means → whether the KRW 3M target holds ultimately depends on long-term ROE staying well above old-cycle levels. The stock has fallen 52% from its peak, implying a long-term ROE of roughly 18.9%; UBS forecasts average ROE of 40.2% for 2027–2031 — the gap is where UBS sees the pricing opportunity.
Content is for reference only, not financial advice.