You open your brokerage account, search for semiconductors, and buy the VanEck Semiconductor ETF under the ticker SMH. The pitch sounds simple enough. You want exposure to artificial intelligence, data centers, and advanced manufacturing without the anxiety of picking individual winners. You assume you bought a diversified basket of twenty-five companies, where a disaster at one firm is neatly offset by gains across the rest.
That is not what you own. You actually bought a vehicle that hands a fifth of its capital to a single company. And it carries a rule that resets that bet every quarter, whatever the news.
At the end of October, Nvidia accounted for over 20% of the entire fund. It was sitting right at the index ceiling. Two weeks later, Jensen Huang resigned after reports about a personal relationship set off a national-security review. The chief financial officer took over as interim chief executive.
One in five dollars of this fund sat in a company whose founder had just walked out the door. Yet by the fund's own rules, the next quarterly rebalance will quietly top that position back up to a fifth of the portfolio. The index does not care that the founder left. It cares that Nvidia is still the biggest company in it.
That is not a risk-management brake. It is a risk-control costume over a fixed target: a fifth of your money stays in the biggest name whether that name is soaring or collapsing. Over the last two years, that rule generated staggering wealth for anyone holding the ticker. Over the last two months, it showed just how violent the ride can be when the market turns.
What you actually own
VanEck likes to say the fund holds twenty-six positions, but the twenty-sixth line is simply cash. You own twenty-five operating businesses. Yet calling this a twenty-five-stock portfolio is mathematically misleading. Because of extreme market-cap weighting, the top ten names account for over 69% of the fund. The top three positions—Nvidia, Taiwan Semiconductor, and Broadcom—command nearly 34% of your money.
Crunch the effective concentration and SMH behaves like an equal-weighted portfolio of about fourteen companies. You are not buying the semiconductor industry. You are buying a mega-cap tech vehicle with a small tail of mid-tier names behind it.
The second surprise is geography. Most investors treat this as an American tech fund. It is not.
Roughly 21% of the fund sits in foreign corporations, held via American depositary receipts. Taiwan Semiconductor accounts for nearly 10% of the portfolio on its own. Dutch lithography giant ASML makes up another 4.5%, while South Korean memory manufacturer SK hynix commands nearly 4%. If export controls tighten or cross-strait tensions flare, more than a fifth of your portfolio is directly in the splash zone.
You are also buying the entire industrial supply chain rather than pure chip designers. Think of the semiconductor world in three layers. At the base are the equipment makers who build the multi-million-dollar tools required to print silicon. That group, led by ASML, Applied Materials, Lam Research, and KLA, takes up roughly 18% of the fund. In the middle sits design software, with Cadence and Synopsys taking another 5%. Only then do you reach the designers and fabricators like Nvidia, AMD, and Texas Instruments.
When you buy SMH, you are not merely betting that tech giants will keep designing chips. You are betting that the factories buying extreme ultraviolet lasers and the software engineers running logic simulations will stay equally flush with cash.
How the sausage is made
To understand why the fund behaves this way, you have to look at the machinery running behind the curtain. SMH is a passively managed exchange-traded fund that tracks the MVIS US Listed Semiconductor 25 Index. That benchmark is administered by MarketVector Indexes, a firm operating directly within VanEck's own corporate family.
The rules look simple. A company must earn at least half its revenue from chips or chip-making equipment, and it must be listed in the US. The fifty largest eligible names are ranked on size and three-month trading volume. The top ten go in automatically.
Here is the first structural wrinkle: the remaining fifteen spots are cushioned by an incumbent buffer. The index rules allow existing constituents ranked anywhere from eleven to forty to keep their seats. That cushion keeps portfolio turnover low—historically around 12% a year—but it quietly favors entrenched market incumbents over fast-rising challengers.
The real fireworks happen inside the weighting formula. The index splits constituents into two buckets. The first group holds between five and ten mega-cap names, capped at 50% of the entire fund. Within that group, every stock is capped at a 20% weight and given a 5% floor. Every company outside that elite group is capped at 4.5%.
The crucial catch is that these caps bind only on the quarterly rebalance date, which lands on the third Friday of March, June, September, and December. In between, weights drift wherever the market takes them. Run hard and a name's slice grows past the ceiling. Fall and it shrinks below.
On rebalancing day the index resets the biggest names to target. Trim back to 20% if the stock ran, top it back up to 20% if it sank. A fifth of the fund is the answer the rule gives for the largest stock in the index, no matter what just happened to it.
The cap is not a judgment about the stock. It is a fixed target.
The mechanics of the wrapper itself are clean. SMH uses 100% in-kind creation and redemption units of 50,000 shares, supported by thirty-eight authorized participants. When big institutional investors want to move money in or out, they swap physical baskets of stock rather than cash. That keeps capital-gains taxes almost entirely off your annual tax return.
The bill
Owning SMH costs you 0.35% a year in management fees. VanEck operates an expense-payment arrangement where the adviser covers fund overhead out of that fee, leaving the gross and net expense ratios identical.
What does 0.35% actually mean for your wallet? On a $10,000 investment growing at an assumed 8% gross annual rate over ten years, that fee will consume roughly $690 of your ending balance. On the fund's current asset base of roughly $60.5 billion, that expense ratio translates into roughly $212 million in annual fees collected by the issuer.
Compare that to its primary rival, the iShares Semiconductor ETF, known by its ticker SOXX. The iShares fund charges roughly 0.33%. Over ten years on that same $10,000 stake, the fee difference between the two amounts to about $39. That is essentially lunch money. Anyone choosing between these two funds based on headline expense ratios is looking at the wrong variable. The difference lies in the weighting caps, not the fee table.
Trading friction is minimal. Thanks to average daily volume near eight million shares and over $4 billion changing hands daily, the median bid-ask spread sits at a razor-thin 0.02%.
The fund does engage in securities lending, loaning out roughly 0.11% of its portfolio to generate extra revenue. That practice introduces an element of institutional counterparty risk, but it also helps offset tracking costs. Over the past decade, SMH has produced an annualized tracking difference of just negative 0.11 points against its benchmark index. In plain English, the fund has handed investors the gross index return minus the management fee, with virtually zero operational leakage.
The scoreboard
The long-term performance numbers look like a misprint from a speculative bull market. Since its inception in late 2011, SMH has generated an annualized net total return of 28.39%. Over the trailing ten-year period through October, it compounded at 32.60% a year. In the three years through October, it delivered an eye-popping 56.20% annualized.
Those headline numbers conceal a brutal reality check that arrived this autumn. At the end of September the fund was sitting on a year-to-date gain of nearly 69%. A month later that was 45%. By mid-November it was 27.8%, with the fund roughly 31% below its June peak. Roughly 41 percentage points of gain disappeared in seven weeks.
This brings us to the showdown with its main rival. Over the trailing decade, measured on market price through September, SMH compounded at roughly 34.4% a year against 32.4% for the iShares product. SMH has also beaten SOXX in nine out of fifteen calendar years.
Yet the honest truth behind that victory is uncomfortable. SOXX was actually ahead of SMH at every single calendar year-end from 2013 through 2023. The entire lifetime margin of victory for SMH was minted during the manic artificial intelligence run of 2024 and 2025.
The reason comes down to diversification. SOXX tracks a different benchmark that caps its top five positions at 8% and the rest at 4%. Its largest holding, AMD, sits near 10%, while Nvidia accounts for less than 8%. When Nvidia went parabolic, SMH rode that single rocket ship to market stardom while SOXX was held back by its stricter caps. When leadership in the chip sector rotated into other names late this year, SOXX came out ahead.
The valuation picture has inverted in a way that catches most growth investors off guard. At the end of October, SMH carried a trailing price-to-earnings ratio of 27.56 and a price-to-book multiple near 9.8. Yet on forward earnings estimates, the fund now screens anywhere between the mid-teens and 21 times, depending on which data provider you consult.
Consider what that means. The forward price-to-earnings multiple of the broad S&P 500 index recently sat at 17.7. For the first time in recent memory, the biggest and most concentrated semiconductor fund on the market is trading at or below the multiple of the broader stock market.
Do not mistake that compressed multiple for automatic safety. The S&P 500 forward multiple contracted this year largely because forward corporate earnings estimates rose by more than 33%, not because equity prices collapsed. In cyclical industries like semiconductors, forward earnings estimates are notoriously fragile. When demand softens, earnings do not gently decline; they fall off a cliff. Buying cyclical chip stocks at a low multiple on peak-cycle earnings is the oldest value trap in modern finance.
What you are really betting on — and the bear case
Own SMH today and you are making one specific bet. Global spending on AI data centers keeps expanding without a pause. You are betting the biggest tech companies keep spending tens of billions a quarter on advanced graphics chips, memory and networking gear.
A short seller would put it more bluntly.
Start with the structural fault line. More than 27% of your capital is parked in just two corporations. The single largest holding, representing over 17% of the portfolio, just suffered the sudden departure of its legendary founder and chief executive. The company is currently operating under an interim chief executive who is simultaneously serving as the chief financial officer. No matter how competent she is, running both jobs while navigating a federal national security inquiry is an enormous managerial distraction.
Meanwhile, the rebalancing formula means you cannot get off. Every three months the biggest name is reset to a fifth of the fund, bought when it has fallen and trimmed when it has run. That is fine while the story holds. When it breaks, a 20% anchor pulls the whole portfolio down with it.
History provides the receipts. This fund does not gently correct; it crashes. Over the past eight years, SMH has suffered four separate drawdowns greater than 28%. It fell 28.4% in 2018, and it plunged 45.3% during the tech wreck of 2022.
It then lost 36% from mid-2024 into early 2025. And between its June peak and its November trough this year, it dropped 33.1%.
The flash crash on November 5 demonstrated how fragile this market structure has become. SMH plunged 11.2% in a single trading session, marking its worst day since the onset of the pandemic. Micron fell 17%, the equipment makers dropped 14%, and Nvidia lost more than 9%.
The catalyst was political and physical. After the midterm elections, ballot measures in several states signaled local resistance to the power and water demands of AI data centers. UBS was reported estimating that roughly 30% of 2028 data-centre capacity is now at risk. If the buildings do not get built, the servers inside them never get ordered.
Add the violent cyclicality of memory chips. Micron and SK hynix account for more than 8% of the fund. We have already seen what happens when that niche cracks: Roundhill's specialized memory fund has lost more than half its value since the summer. Factor in tightening American export licensing rules on the 10% of the fund based in Taiwan, and you have a powder keg of cyclical, geopolitical, and structural risks.
What would prove the bears entirely right? Watch two specific signals. The first is a second consecutive quarter of capital expenditure reductions from the mega-cap cloud providers. The second is Nvidia's new leadership guiding gross margins below the company's long-run range on the first call that is genuinely theirs. If either event materializes, the earnings denominator will collapse, and this fund will re-rate downward with astonishing speed.
How to use it
Given that risk profile, SMH has no business serving as a core foundation for an investment portfolio.
SMH belongs exclusively in the satellite sleeve of an aggressive equity portfolio, capped at roughly 5% to 10% of total assets. It is designed to replace individual stock picking for investors who want focused upside exposure to the semiconductor cycle. If you already own heavy positions in broad-market tech funds, adding SMH simply doubles down on the exact same mega-cap names you already own.
It suits young, risk-tolerant investors who can watch their capital fall by 40% in a year without panic-selling. It also suits traders who want a liquid, low-spread way to express a view on chip demand.
It is wrong for conservative savers, for retirees drawing income, and for anyone who needs the money back soon. The fund yields 0.17% on a 30-day SEC basis, pays its small dividend once a year, and can drop 30% in weeks.
The verdict
VanEck Semiconductor ETF is not an all-weather vehicle. It is a high-octane momentum tool engineered to deliver maximum exposure to the sharpest edge of the technology cycle.
If you choose to own it, do so with your eyes wide open. You are not buying twenty-five independent companies. You are buying a top-heavy basket where three names dictate your returns, wrapped around a rigid rule that resets a fifth of your money into one stock every quarter.
For the next several months, watch the December rebalance and the capital-spending commentary from the big cloud providers. If those budgets hold, this fund keeps compounding at rates few products can match. If they get trimmed, that 20% weight becomes the anchor.
Analysis, not advice. We may own these names and we may sell them at any time.




