Citi: Buy KRW on Dips at 1400, Targeting 1330 by Year-End
nashnova research
Citi Korea head of trading Lee Sanghun calls USD/KRW 1,400 a contrarian entry to buy the won, targeting 1,330 by year-end; the bull case rests not on central-bank intervention but on a mounting dollar supply from memory-chip exports.
The won hit 1,400 — why does Citi say buy?
Lee identifies the Fed's hawkish tilt and rising oil prices as the main short-term drags, potentially pushing USD/KRW past the 1,400 handle.
But he frames this as a contrarian entry point, not a bearish trend signal. This means → Citi treats the near-term headwinds as a discount window, not a risk escalation.
The won fell for a fourth straight session Thursday, touching 1,388.25; it steadied near 1,381 at Friday's open.
What underpins the recovery call — how strong are chip exports?
Korea's August exports rose 68.7% year-on-year to $98.26 billion; semiconductor exports alone hit $46.65 billion, producing a monthly trade surplus of $34.75 billion.
In plain terms = Korea is earning a growing pile of dollars from chip sales, and those dollars eventually need converting into won to pay wages, taxes, and operating costs — the won's "bid" is embedded in the export data.
Lee draws a sharp line between the won and the yen: the yen's rise depends on Bank of Japan rate normalization and intervention, while the won's support comes from real FX income generated by the semiconductor supply chain. This reflects two fundamentally different currency stories.
If exports are this strong, why is the won still falling?
Lee notes that exporters were "too eager to sell dollars" in July and August; won liquidity is now ample and corporates are "in no rush" to convert further.
Some speculative players are shorting the won in the gap. Put simply = the chip-export dollars are real, but corporates are sitting on them for now — giving speculators room to lean on the won. Strong fundamentals and a short-term pullback can coexist.
This means → the won's near-term weakness is not a breakdown in the export thesis; it is a timing mismatch in conversion flows.
Can the memory giants' earnings sustain this story?
SK Hynix is up 168.05% year-to-date; Q2 operating profit hit roughly ₩60.54 trillion, up 557% year-on-year. High Bandwidth Memory (HBM — high-speed memory designed for AI training), AI-server DRAM, and enterprise SSDs (eSSD) are the profit drivers.
Samsung Electronics is up 110.59% YTD with Q2 group operating profit of ₩89.5 trillion; its memory unit posted record quarterly revenue and profit and is ramping HBM4 sales.
This means → memory chips are not just shipping in volume — unit prices and margins are surging together, and that is the hardest card in the won-recovery hand.
How long can the pricing cycle last — is the medium-term target realistic?
TrendForce projects cumulative 2026 server-DRAM contract-price gains of roughly 270% and enterprise-SSD gains of roughly 235%; 2027 HBM contract prices could still rise 70–140%.
Lee argues that if the semiconductor cycle holds, exporters resume large-scale dollar selling, and Middle East tensions ease, the won could reach the upper end of the 1,200-range in H1 2027.
In plain terms = Citi's year-end 1,330 target is only the first stop; whether the won can push further into the 1,200s depends on whether chip-price expectations actually materialize — that is the key verification node for the entire thesis.
市场有风险，内容仅供研究参考，不构成投资建议。