Quick answer (June 2026): Korean AI semiconductor ETFs are not interchangeable. Over the last 12 months, KODEX AI Semiconductor Core Equipment returned 248.82%, SOL AI Semiconductor Materials/Parts/Equipment returned 201.33%, and TIGER AI Semiconductor Core Process returned 155.59%, according to FunETF and issuer data as of 2026-06-11/12. The reason is simple: Korean funds are not primarily NVIDIA funds. They are HBM, packaging, testing, substrate, and semiconductor-equipment supply-chain funds. If you read our previous US AI & Semiconductor ETFs Complete Guide, this article is the Korean counterpart: same theme, different market structure, different winners.
1. Why Korean AI semiconductor ETFs matter now
In 2026, the Korean semiconductor equity story is no longer just “Samsung versus SK Hynix.” The market has moved deeper into the value chain. High-bandwidth memory (HBM) capacity expansion, advanced packaging, testing, and substrate suppliers have outperformed many of the memory makers themselves. That is why Korean semiconductor ETFs have diverged sharply even though they all carry similar names. US funds such as SMH and SOXX ride the earnings power of NVIDIA, Broadcom, TSMC, and AMD. Korean funds ride the suppliers that make AI memory scaling possible: Hanmi Semiconductor, Leeno Industrial, Isu Petasys, Doosan, and Jusung Engineering.
This difference is structural, not cosmetic. Korea is one of the world’s most important memory-manufacturing hubs. So domestic semiconductor ETFs naturally lean toward the infrastructure of AI chips rather than the fabless design layer. For investors, that means Korean ETFs are closer to “AI infrastructure picks and shovels” than to “AI chip platform monopolies.”
2. Snapshot table — issuer, cost, size, profile
| ETF | Issuer | Headline fee | Real cost ratio | AUM | Listed | Profile |
|---|---|---|---|---|---|---|
| KODEX AI Semiconductor Core Equipment | Samsung Asset Management | 0.39% | 0.5270% | KRW 525.60bn | 2023-11-21 | Equipment / backend concentration |
| TIGER AI Semiconductor Core Process | Mirae Asset | 0.45% | 0.6578% | KRW 197.84bn | 2023-11-21 | HBM / process concentration |
| SOL AI Semiconductor MPE | Shinhan Asset Management | 0.45% | 0.5208% | KRW 1.3349tn | 2023-04-25 | Broad materials, parts & equipment |
The size gap is the first thing to notice. SOL is the category heavyweight at KRW 1.3349tn, roughly 2.5 times the size of KODEX and about 6.7 times TIGER. That matters for liquidity, market impact, and position sizing. KODEX, however, offers the sharpest equipment tilt, while TIGER is the narrowest and most explicitly tied to HBM-process beneficiaries.
3. The cost trap — total fee vs real all-in cost
Korean ETFs disclose more than the headline management fee. Investors should look at headline fee, TER, real cost ratio, and premium/discount behavior together.
| ETF | Headline fee | TER | Real cost ratio | 3M premium/discount | Takeaway |
|---|---|---|---|---|---|
| KODEX | 0.39% | 0.44% | 0.5270% | -0.0818% | Lowest cost in the group |
| TIGER | 0.45% | 0.51% | 0.6578% | -0.0801% | Highest all-in cost |
| SOL | 0.45% | 0.51% | 0.5208% | 0.0403% | Efficient despite larger size |
The important surprise is that SOL’s real cost ratio is slightly lower than KODEX’s in practice-adjusted terms of TER-plus-market behavior, despite the same 0.45% headline fee as TIGER. Meanwhile TIGER is meaningfully more expensive once the real cost ratio rises to 0.6578%. On a long hold, that difference compounds.
4. Holdings concentration — this is where the products truly diverge
This is the most important section. These are not three generic Korea semiconductor funds. They are three very different expressions of the same macro theme.
| ETF | Top holding | 2nd | 3rd | Top 3 total | Interpretation |
|---|---|---|---|---|---|
| KODEX | Hanmi Semiconductor 26.08% | Doosan 20.74% | Leeno Industrial 9.26% | 56.08% | Ultra-concentrated equipment/backend bet |
| TIGER | Hanmi Semiconductor 24.45% | Leeno Industrial 16.96% | Isu Petasys 15.06% | 56.48% | Ultra-concentrated HBM/process bet |
| SOL | Hanmi Semiconductor 18.11% | Jusung Engineering 12.23% | Isu Petasys 9.12% | 39.46% | Broader supply-chain diversification |
KODEX and TIGER both have more than 56% of assets in just three names. That is concentration on a level comparable to, and in some respects more severe than, a mega-cap-heavy US semiconductor ETF. SOL is still thematic and aggressive, but its top-three sum of 39.46% is materially more diversified. In practical portfolio terms:
- KODEX is the cleanest way to bet on Korean semiconductor equipment and backend names.
- TIGER is the most explicitly tied to HBM process leverage, testing, and substrate spillovers.
- SOL is the best core holding if you want supply-chain breadth rather than a narrow tactical punch.
This is also the biggest difference from the US guide. In our previous US article — US AI & Semiconductor ETFs Complete Guide — SMH was effectively a NVIDIA + TSMC fund, while SOXX was the more diversified pure-semi option. In Korea, the dividing line is not “NVIDIA vs diversified semis.” It is “backend/process concentration vs broader materials/parts/equipment spread.”
5. Performance by time horizon
| ETF | 1M | 3M | 6M | 1Y | YTD |
|---|---|---|---|---|---|
| KODEX | 3.24% | 31.06% | 108.11% | 248.82% | 113.83% |
| TIGER | 3.88% | 15.16% | 100.13% | 155.59% | 93.27% |
| SOL | 2.73%-8.82% | 30.24%-32.72% | 102.59% | 201.33%-207.56% | 103.90% |
Over one year, KODEX is the clear winner at 248.82%. SOL sits in the middle at just over 200%, and TIGER trails at 155.59% despite still being an extraordinary performer in absolute terms. Over three months, however, KODEX and SOL are much closer to each other, while TIGER has lagged. That tells you something important: the 2025-2026 supercycle rewarded equipment and broad supply-chain winners more consistently than the narrowest HBM-process basket.
Compare that with the US market. In the US guide, SMH and SOXQ rode the GPU-memory-networking complex through NVIDIA, TSMC, Micron, and Marvell. In Korea, the equivalent outperformance came through Hanmi Semiconductor, Doosan, Jusung Engineering, Leeno Industrial, and Isu Petasys. The same AI boom produced different stock-market winners because each market occupies a different layer of the AI stack.
6. Risk analysis — delisting, liquidity, concentration, and no direct FX hedge issue
- Concentration risk: KODEX and TIGER have top-three weights above 56%, so single-name setbacks matter a lot.
- Theme volatility: When a theme posts 1-year returns between 155% and 249%, expectations are already elevated. Any slowdown in HBM capex or AI-server demand can hit hard.
- Liquidity: SOL leads with about KRW 391.93bn in average traded value, versus KRW 92.03bn for KODEX and KRW 63.49bn for TIGER.
- FX exposure: These are KRW-traded domestic ETFs, so investors do not carry the same direct USD/KRW translation risk they would with SOXX or SMH. But earnings sensitivity remains globally linked.
- Delisting risk: All three are large enough that near-term delisting risk is low, especially SOL at over KRW 1.3tn in assets.
7. Korea vs the US — how they differ from SOXX and SMH
| Axis | US SOXX / SMH | Korean KODEX / TIGER / SOL |
|---|---|---|
| Main exposure | AI chip design, foundry, fabless scale | HBM, packaging, testing, materials/equipment |
| Representative names | NVIDIA, TSMC, Broadcom, AMD | Hanmi Semi, Leeno, Isu Petasys, Doosan, Jusung |
| FX risk | Direct USD exposure | KRW-listed, lower direct FX risk |
| Concentration profile | SMH concentrated, SOXX diversified | KODEX/TIGER concentrated, SOL more diversified |
| Investment logic | Own the AI chip profit pool | Own the AI memory and manufacturing supply chain |
The best way to think about this is simple: US semiconductor ETFs buy the brains of AI; Korean semiconductor ETFs buy the industrial plumbing that lets those brains scale. They are complements, not substitutes.
8. Recommendations by investor type
- Aggressive: KODEX. Best 1-year return, lowest headline fee, and the purest equipment/backend tilt.
- Balanced/core: SOL. Largest AUM, best liquidity, and the broadest diversification in the group.
- High-conviction tactical: TIGER. Best for investors who specifically want HBM/process leverage and can tolerate higher cost and concentration risk.
If I had to frame a practical allocation in mid-2026, I would favor SOL as the core holding and KODEX or TIGER as the satellite position. The theme has already rallied too far for blind lump-sum buying to look prudent. A staged entry over several months makes more sense.
9. FAQ
Q1. Which Korean AI semiconductor ETF is best for a one-fund core position?
SOL is the easiest core choice because it has the largest asset base, the best liquidity, and the least extreme concentration of the three.
Q2. Why isn’t KODEX automatically the best just because it has the strongest 1-year return?
Because 46%+ of the fund is basically Hanmi Semiconductor plus Doosan. That concentration is a strength on the way up and a vulnerability on the way down.
Q3. Can investors hold Korean ETFs together with SOXX or SMH?
Yes. In fact, the combination can make sense because the exposures sit on different layers of the AI value chain: US design/fabless leaders versus Korean HBM/process/equipment suppliers.
10. Related reading
- US AI & Semiconductor ETFs Complete Guide (2026)
- US Quantum Computing ETFs
- Korean Quantum Computing ETFs Comparison
Source: https://www.samsungfund.com/etf/product/view.do?id=2ETFL7