KOSPI’s Two Semiconductor Titans: Why Samsung and SK hynix Still Look Undervalued in the AI Memory Era
Disclaimer: This article is for informational and educational purposes only and does not constitute investment advice, a recommendation, or an offer to buy or sell any security. Investing involves risk, including loss of principal and foreign-exchange risk.
3-Line Summary
KOSPI offers a rare mix: world-class semiconductor exposure plus a market that has historically traded at a persistent valuation discount.
The cleanest, most concentrated way to express that thesis is Samsung Electronics (KRX: 005930 / 005935) and SK hynix (KRX: 000660).
In my bull case, Samsung is “open to” ~$200 and SK hynix to ~$1,000 on a USD-equivalent basis—if AI-memory demand stays strong and Korea’s discount compresses.
Abstract
This guide outlines a thesis-driven approach for English-speaking investors seeking exposure to Korea’s equity market through two dominant semiconductor companies: Samsung Electronics and SK hynix. The argument is that both names have experienced periods of deep undervaluation relative to their strategic positioning, and that a multi-year AI-driven memory cycle can support sustained earnings power. I present a numerical bridge from today’s prices (whatever they are when you read this) to USD-equivalent ceilings—explicitly modeling FX translation, earnings normalization, and valuation multiple changes. This is a scenario analysis, not a promise.
1. Why KOSPI, and why now?
Korea is not a “small market.” It is one of the most important nodes in the global electronics supply chain. Yet KOSPI has often traded with a structural discount—commonly explained by a mix of governance concerns, capital allocation skepticism, and investor preference for U.S. mega-cap liquidity.
The opportunity in 2026 and beyond is that two things can happen simultaneously:
Earnings power improves (AI-related demand lifts the memory and advanced packaging stack).
The discount compresses (even modestly), meaning investors pay a higher multiple for the same earnings.
When both happen, equity prices can move much further than “earnings growth alone” would suggest.
2. The simplest instruments: what to buy
2.1 Direct KRX exposure (most precise)
Samsung Electronics common: KRX 005930
Samsung Electronics preferred: KRX 005935
SK hynix: KRX 000660
Preferred shares can sometimes trade at a discount to common shares while offering similar economic exposure (but with different voting rights and liquidity characteristics). If you’re building a long-term position, the common vs. preferred decision is often about pricing efficiency + liquidity, not “business fundamentals.”
2.2 Single-ticket Korea exposure (less precise, but easy)
A broad Korea ETF can be convenient, but you should understand a structural fact: Korea index exposure is often heavily driven by these two semiconductor names anyway. If your thesis is primarily “AI-memory + Korea discount compression,” direct ownership is the cleanest expression.
3. The core thesis in one sentence
Samsung + SK hynix are leveraged to an AI-driven memory upcycle, and KOSPI-level valuation discount compression can amplify returns beyond what the earnings cycle alone would imply.
This thesis has two engines:
Engine A: Semiconductor cycle upgrades (HBM, high-end DRAM, advanced packaging constraints).
Engine B: Multiple expansion (a “less discounted Korea” narrative becomes more credible).
4. Why AI makes memory different (not risk-free—different)
4.1 AI accelerators are memory-hungry
Modern AI systems don’t just need compute; they need fast, high-bandwidth memory close to the accelerator. The practical consequence is that premium memory (especially HBM-class products) becomes a bottleneck and captures higher value.
A conservative, investor-friendly way to think about it:
The AI stack increases memory per unit of compute.
It increases the importance of bandwidth and packaging, not just raw capacity.
It raises the odds of longer-than-usual tight supply in premium segments.
4.2 The cycle still exists
Memory remains cyclical. Even in a strong secular trend, demand arrives in waves:
AI capex can surge for 4–8 quarters, then “digest” for 2–4 quarters.
Pricing can stay strong longer than skeptics expect, then drop faster than optimists expect.
This is why position sizing and staggered entries matter.
5. Company theses
5.1 Samsung Electronics: diversified giant, memory torque
Samsung is a “system” company:
Memory (DRAM/NAND) provides operating leverage in upcycles.
Foundry/logic adds strategic optionality, but competition and capital intensity are real.
Consumer electronics provides scale and brand, but can soften during global slowdowns.
Why undervaluation can be extreme:
In downcycles, the market often prices Samsung as if memory weakness is permanent. When the cycle turns, earnings can rebound sharply, and the market re-rates the stock quickly—especially if the starting valuation is depressed.
5.2 SK hynix: more concentrated, more AI-memory leverage
SK hynix is more directly tied to:
high-end DRAM,
premium memory mix,
AI-related demand dynamics.
Why upside can be cleaner:
If premium memory demand remains strong, SK hynix can show earnings acceleration that makes “big price numbers” mathematically plausible without needing a ridiculous valuation multiple.
6. Turning your USD ceilings into hard numbers (FX translation)
Your ceilings are:
Samsung ≈ $200 (USD-equivalent)
SK hynix ≈ $1,000 (USD-equivalent)
For a USD-based investor, you must translate those into KRW price levels. The formula is:
KRW Target = USD Target × (KRW per USD)
6.1 FX sensitivity table (the most important table in this post)
| FX (KRW per USD) | Samsung $200 → KRW target | SK hynix $1,000 → KRW target |
|---|---|---|
| 1,200 | ₩240,000 | ₩1,200,000 |
| 1,300 | ₩260,000 | ₩1,300,000 |
| 1,400 | ₩280,000 | ₩1,400,000 |
| 1,500 | ₩300,000 | ₩1,500,000 |
| 1,600 | ₩320,000 | ₩1,600,000 |
Interpretation:
If KRW strengthens (lower number), the USD ceiling becomes easier to reach.
If KRW weakens (higher number), the KRW price must rise more to hit the same USD figure.
7. The valuation bridge: how do you get there without fantasy multiples?
The cleanest valuation identity for equities is:
Price = Earnings × Multiple
So the question becomes:
How much of the move comes from earnings growth, and how much from multiple expansion?
7.1 A simple “path-to-ceiling” decomposition
Let:
Earnings grow by G (e.g., 1.5×, 2.0×)
Multiple changes by M (e.g., 0.9× contraction, 1.2× expansion)
Then:
Price change ≈ G × M
Here are realistic ranges that can produce large outcomes:
Scenario set (illustrative, not forecasts)
Base cycle rebound: Earnings 1.5×, multiple 1.05× → ~1.58× price
Strong AI cycle + mild re-rating: Earnings 2.0×, multiple 1.15× → ~2.30× price
Bull case (earnings durability recognized): Earnings 2.2×, multiple 1.25× → ~2.75× price
This is the logic behind saying “the ceiling is open.” It’s not “magic.” It’s math—conditional on earnings staying strong and the market paying a less pessimistic multiple.
7.2 Why Samsung can justify ~$200 USD-equivalent
Samsung’s ceiling case usually requires two things:
A meaningful earnings rebound from trough conditions (memory cycle + mix improvement)
A modest re-rating from “permanent discount” toward “normal discounted global tech”
Samsung doesn’t need to trade like a U.S. software stock. It just needs:
a credible multi-quarter earnings run-rate, and
a market that stops pricing the business as structurally stagnant.
7.3 Why SK hynix can justify ~$1,000 USD-equivalent
SK hynix’s ceiling case can be even more straightforward:
If premium memory earnings surge and remain durable for multiple quarters,
the stock can reprice to reflect a longer runway,
and the multiple required can still remain in “semiconductor-normal” territory (often single digits to low teens depending on cycle confidence).
In plain terms: SK hynix can get there with more earnings doing the heavy lifting and less reliance on narrative.
8. Risk model (with numbers you can actually use)
8.1 The three largest risks
Memory pricing rollover: drawdowns of 30–60% are historically possible in semis during hard turns.
AI capex digestion: a 2–4 quarter pause can break momentum and compress multiples.
FX drag: a 10–15% KRW weakening can shave a similar magnitude off USD returns even if KRW stock prices rise.
8.2 A quick USD stress test
Assume the stock rises +60% in KRW, but KRW weakens -12% vs USD.
A rough approximation is:
USD return ≈ (1.60 × 0.88) − 1 = +40.8%
That’s still positive, but meaningfully less than the KRW chart suggests. This is why you should never ignore FX.
9. Implementation: how I would structure the trade (without turning it into a gamble)
9.1 Position sizing (portfolio-level)
A practical range for many diversified investors:
Conservative: 2%–4% total portfolio in Korea semiconductors
Moderate: 4%–8%
Aggressive: 8%–12% (only if you tolerate deep drawdowns)
9.2 Entry method: tranches
Rather than one buy:
Split into 4–8 tranches over 8–20 weeks.
This reduces timing regret in a cyclical sector.
9.3 Rebalancing rules (simple and mechanical)
If you hold both names:
Start at 50/50 (balanced exposure)
Rebalance if one grows to ≥60% of the pair or falls to ≤40%
This prevents “one winner becomes the whole portfolio” by accident.
9.4 Time horizon
This is not a “next week” thesis. A realistic thesis horizon is:
12–36 months for cycle + narrative to play out
(Shorter outcomes happen, but the framework is built for quarters, not days.)
10. Final stance
My view is straightforward:
Samsung Electronics and SK hynix were deeply undervalued at points, largely because the market priced them through the lens of a typical memory cycle and a persistent Korea discount.
The AI-driven memory dynamic can extend profitability and improve mix, while governance and capital allocation perception can gradually compress the discount.
Under those conditions, I believe Samsung has room toward ~$200 and SK hynix toward ~$1,000 on a USD-equivalent ceiling basis.
These ceilings are not guarantees. They are conditional outcomes that become plausible when:
premium memory demand stays strong,
earnings durability becomes visible across multiple quarters, and
the market pays a less pessimistic multiple for Korea’s flagship semiconductor exposure.
Disclaimer: This article is for informational and educational purposes only and does not constitute investment advice, a recommendation, or an offer to buy or sell any security. All scenarios are illustrative, not promises. Consider your objectives, time horizon, and risk tolerance, and consult a licensed professional if needed before investing.
Comments
Post a Comment