Not all semiconductor ETFs are equal — it's about what's inside
Written and maintained by ETF Scanner · Report a correction
The label 'semiconductor ETF' is the same, but open them up and the portfolios are often completely different.
NVIDIA weight decides performance
Semiconductor ETFs differ a lot in how much they hold of a leader like NVIDIA. Some concentrate in a few top names; others cap per-stock weight for even spread.
When NVIDIA rips, the concentrated one wins big; when it corrects, it loses big. That one weight difference moves returns by several %p.
Design vs equipment makers
Chips split into design (fabless), manufacturing (foundry), equipment and materials. Some ETFs stick to large design/manufacturing names; others reach into equipment (ASML, AMAT) and the supply chain.
Equipment and materials swing harder with the semi cycle — more upside in rallies, more downside in drops. Decide whether you want exposure to 'end products' or 'the whole process'.
Korea- vs US-listed
Korea-listed semi ETFs lean on Samsung and SK Hynix; US-listed ones center on NVIDIA, Broadcom and TSMC. They barely overlap — completely different bets.
'Investing in semiconductors' is the same phrase, but which market and which stocks you're exposed to is entirely different.
So what to check
Compare holdings, weights and concentration (HHI) instead of the name. Open two semi ETFs side by side in ETF Scanner and the top-weight and overlap jump out. First confirm the concentration is one you can stomach.
Semiconductors are not one industry
Treating chips as a single sector makes choosing impossible. There are at least four layers: firms that only design, foundries that manufacture to others' designs, equipment makers, and wafer and materials suppliers.
They move at different times. Designers lead when demand returns; equipment moves only once capacity is committed; materials track utilisation and lag further. Two 'semiconductor ETFs' react differently to the same headline depending on which layer they weight.
Memory and logic split too. Memory swings on cyclical gluts and shortages; logic runs on long design relationships and moves more gently. Korean large caps skew to memory, the US benchmark to designers.
What geography changes
Korean semiconductor funds usually put more than half their weight in two names. Calling that a bet on two companies is fair. If you bought it expecting diversification, it is not that.
The US benchmark spreads its top weights more and includes designers and equipment together — but adds currency to your return. Without checking whether the fund hedges, you cannot explain why the index rose more than your account.
Korean-listed versions of overseas semiconductor indices add one more layer: hedged and unhedged share classes track the same index and return differently, so read the suffix.
A three-minute check
Start with the top ten and their combined weight. If two names exceed half the fund, their news is the fund's news.
Then concentration. Thirty holdings still means little diversification if the top is heavy. HHI next to the top-10 weight settles it.
Finally overlap with what you own. If you already hold Korean large caps directly or a KOSPI 200 fund, adding a Korean semiconductor ETF mostly increases the same two positions.
This article is general information, not investment advice. Investment decisions and responsibility are your own.