Banks split on memory: Morgan Stanley slashes SK hynix target 38% as Goldman calls the sector oversold
Morgan Stanley analyst Shawn Kim cut SK hynix's target price 38% to ₩1.6 million and Samsung's 19% to ₩309,000, warning about margins after 2028 — the same day Goldman Sachs called semiconductor stocks oversold and Nomura held targets far above both. The three banks' SK hynix targets now span nearly 3x.
The News
The banks split on memory on Tuesday, and the gap between them is now enormous. Morgan Stanley analyst Shawn Kim cut SK hynix's target price 38%, from ₩2.6 million to ₩1.6 million, and Samsung's 19%, from ₩381,000 to ₩309,000 — the preferred shares, a class of stock with a fixed dividend, were cut from ₩304,800 to ₩247,200. He kept Overweight ratings on both, meaning he still expects the stocks to beat the market.
Kim's case is about what happens after 2028. Rising memory supply from late 2027, plus delays to data-centre construction, could compress long-term margins — the share of revenue a company keeps as profit. When new factories come online and supply grows faster than demand, prices fall and margins shrink. That is the classic memory cycle, and Kim is saying it arrives in 2028. Yet the same note raised 2027 operating-profit forecasts — 9% for Samsung and 13% for SK hynix — so the bank is bullish on next year and cautious about the years after it.
Goldman Sachs read the same market very differently on the same day. It called semiconductor stocks oversold, noting the Philadelphia Semiconductor Index — a basket of the biggest US chip stocks — had fallen 13% in a week and 31% since November 1, with Samsung and SK hynix near 1.21x and 1.23x 12-month forward price-to-book. Price-to-book compares a company's share price to the value of its net assets; near 1.2x is low for companies earning what these two are earning. Nomura, meanwhile, held its targets far above both: ₩4.7 million for SK hynix and ₩670,000 for Samsung. The three banks' SK hynix targets now span nearly three times.
Three banks, three answers — and the SK hynix targets span nearly 3x.
Why It Matters
The split matters because it is a disagreement about the cycle, not about the quarter. Everyone agrees memory is booming now: prices are rising, profits are up, and Seoul's rally says the market believes it. The argument is about what happens when the boom ends. Morgan Stanley says the end comes in 2028, when new supply arrives and data-centre delays bite. Goldman says the sell-off has already gone too far, and the stocks are cheap. Nomura says the targets should be far higher still. The gap between those views — nearly three times on SK hynix — is a measure of how uncertain the cycle's timing really is.
The 2027 numbers are the detail that keeps Morgan Stanley's call honest rather than simply bearish. A bank that cuts a target 38% while raising next year's profit forecast by 13% is not saying the business is broken — it is saying the business is great now and the trouble comes later. That is a harder call to dismiss than a straight downgrade, and it is also a harder call to act on: if the trouble is two years away, the stock can run a long way first.
The valuation numbers are the counterweight. Samsung and SK hynix at about 1.2x forward price-to-book are cheap by the standards of companies whose profits are rising as fast as theirs — and that is Goldman's whole argument. The market has already priced in a lot of fear: the Philadelphia Semiconductor Index is down 31% since November 1. The question for the rest of the year is whether that fear was the overshoot Goldman sees, or the early warning Morgan Stanley sees.
Analyst's Take
Our call is Mixed. The disagreement itself is the story: one bank cuts SK hynix's target 38% while another holds a target nearly three times higher, and a third calls the whole sector oversold. All three cannot be right, and the market's own answer — Seoul rallied for a second straight day — sides with the bulls for now. But Morgan Stanley's warning is about 2028, not next week, and a 2028 supply glut is exactly the kind of slow-moving risk that rallies ignore until they cannot.
What to watch: memory contract prices in the fourth quarter, and the data-centre construction pipeline. If prices keep rising and the building delays clear, Morgan Stanley's 2028 case weakens. If supply starts arriving faster than demand, the cut targets start to look like the honest ones.
Are you bullish or bearish on this story?
By The Numbers
FAQ
What is a target price?
A target price is an analyst's estimate of what a stock will be worth in about a year. Morgan Stanley cut SK hynix's target 38%, from ₩2.6 million to ₩1.6 million, and Samsung's 19%, from ₩381,000 to ₩309,000 — but kept Overweight ratings on both, meaning it still expects them to beat the market.
Why did Morgan Stanley cut the targets?
Analyst Shawn Kim's case is about what happens after 2028: rising memory supply from late 2027 and delays to data-centre construction could squeeze long-term profit margins. A margin is the share of revenue a company keeps as profit — when margins compress, each sale earns less.
Did Morgan Stanley turn negative on memory?
No — and that is the twist. The same note raised 2027 operating-profit forecasts by 9% for Samsung and 13% for SK hynix. The bank is bullish on next year and cautious about the years after it.
What did Goldman Sachs say?
Goldman called semiconductor stocks oversold the same day, noting the Philadelphia Semiconductor Index had fallen 13% in a week and 31% since November 1, with Samsung and SK hynix near 1.21x and 1.23x 12-month forward price-to-book — a low valuation measure that compares the share price to the company's net assets.
How far apart are the banks?
Very far. Nomura kept its targets at ₩4.7 million for SK hynix and ₩670,000 for Samsung — nearly three times Morgan Stanley's cut targets. The three banks' SK hynix targets now span nearly 3x.
👁
Source: MoneyToday