Rising AI Demand Shifts Plans for SK Siltron Sale
0xBroomberg
SK Group's silicon-wafer subsidiary SK Siltron has seen its valuation surge from ₩5 trillion to ₩7 trillion as AI chip demand explodes, forcing the conglomerate to reconsider selling — what was set to be one of Korea's biggest 2026 deals may not happen at all.
What is SK Siltron, and why is it suddenly worth more?
SK Siltron is one of the world's major suppliers of silicon wafers — the base material chips are built on, essentially "flour for semiconductors." SK Group holds roughly 51%.
When listed for sale in 2025, the company was valued at about ₩5 trillion. With AI infrastructure spending surging, the market has recently repriced it at roughly ₩7 trillion — a 40% jump in under a year.
This means → the gap between buyer and seller expectations is widening, making a deal that once looked close harder to close.
Why have talks with Doosan stalled?
SK Group picked Doosan as the preferred buyer in late 2025 and came close to signing a share-purchase agreement (SPA), but progress has slowed markedly since.
The core obstacle is a valuation gap: SK Siltron keeps getting more valuable, and Doosan's offer hasn't kept pace.
The market has a second concern: can Doosan's credit rating and financial firepower match SK Group's ability to support the business through a downturn? In plain terms = if the semiconductor cycle turns cold, can Doosan absorb the hit?
Why does SK Hynix want to keep it in-house?
SK Hynix reportedly submitted a report to SK Group leadership recommending a reassessment of SK Siltron's strategic value.
The backdrop: rapid growth in AI memory, HBM (high-bandwidth memory), and advanced DRAM is making supply-chain stability for premium silicon wafers increasingly critical for SK Hynix.
This means → for SK Hynix, keeping its upstream wafer supplier in-house is worth more than the cash from a sale — especially as the AI arms race accelerates.
Has SK Group's financial pressure eased?
SK Group originally pushed the sale to improve its balance sheet, but the picture has changed.
Net borrowings fell from ₩82 trillion at end-2023 to ₩38.2 trillion at end-2025 — a drop of more than half.
SK Hynix, riding HBM and AI memory demand, posted record earnings, holds substantial cash-like assets, and has raised capital via ADRs (American Depositary Receipts) in the U.S. This reflects a disappearing premise: the group no longer needs to sell under financial duress.
Has Chairman Chey's personal calculus changed?
Another driver behind the sale was SK Group Chairman Chey Tae-Won's divorce proceedings — the market expected him to sell his personal 29.4% stake to fund a potentially massive property settlement.
But with a retrial verdict approaching and group finances improving, whether he still needs to liquidate immediately is being reassessed.
In plain terms = the original logic was "short on cash and facing a divorce payout." Now cash is less tight, and the divorce ruling isn't final — the urgency to sell has weakened.
Where does this end — and what does each path mean?
Path one: sell to Doosan → the valuation gap and financial-capacity concerns are hard obstacles; unless Doosan raises its bid substantially, this path is difficult.
Path two: fold into SK Hynix → Chey has said he wants to double semiconductor capacity within five years; keeping SK Siltron gives SK Hynix tighter control over wafer supply, quality, and materials R&D.
Path three: status quo → neither sell nor merge, wait for the retrial verdict and clearer market conditions. This means → the deal's fate hinges on a court ruling and SK Group's ultimate call on supply-chain strategy in the AI era.
Content is for reference only, not financial advice.