KEY POINTS
- The VanEck Semiconductor ETF (SMH) has rallied roughly 20% since April 1 and has pulled in $7.3 billion in net inflows over the past year, nearly four times what SOXX absorbed.
- SMH now holds roughly $50 billion in AUM, with Nvidia and TSMC combined accounting for more than 30% of the portfolio.
- The next catalyst is Nvidia's fiscal Q1 earnings on May 27; an in-line print on data center revenue sustains the SMH bid, a miss forces the rotation narrative.
The VanEck Semiconductor ETF has rallied roughly 20% since April 1 and has pulled in $7.3 billion in net inflows over the trailing 12 months — nearly four times the $1.9 billion absorbed by the competing iShares Semiconductor ETF (SOXX) over the same period. SMH now holds approximately $50 billion in AUM and has become the institutional vehicle of choice for U.S. investors seeking single-ticker exposure to the AI chip buildout, particularly following TSMC's 58% Q1 profit jump and the Q2 revenue guide of $39.0–$40.2 billion. Over the past year, as Nvidia's market cap climbed toward roughly $4.5 trillion and TSMC approached the $1.5 trillion threshold, SMH's concentrated-weight structure delivered returns that few broad-market ETFs came close to matching.
Why SMH Is Winning the Flows Race
The structural difference between SMH and SOXX is simple but consequential. SMH holds 25 names with a modified market-cap methodology that tilts heavily toward the largest pure-play semiconductor companies. Nvidia sits at the cap weight, and TSMC holds the second-largest position at roughly 11%. Together those two names represent more than 30% of the fund. SOXX uses a different methodology — the PHLX Semiconductor Sector Index — which produces a more diversified 30-name basket with lower individual position sizes.
For an allocator who wants pure AI chip beta, SMH's concentration is the feature. For an allocator who wants semiconductor exposure with reduced single-name risk, SOXX is the cleaner vehicle. In practice, flows have voted decisively for concentration. SMH's flows-to-AUM ratio ranks in the top tier of all equity ETFs year-to-date. SOXX is solid — its flows-to-AUM YTD ratio reads 6.63% and ranks in the top 50% — but it is nowhere near SMH's velocity.
The expense ratio gap matters too. Both funds carry expense ratios in the same general range, so the flow decision is not primarily cost-driven. It is a statement of preferred exposure. Institutional allocators added to SMH through the U.S.-Iran volatility window, which is unusual. Normally, geopolitical risk windows see profit-taking in concentrated thematic products. Instead, SMH saw continued net adds, a tell that the AI capex narrative has transitioned from trade to allocation.
The Concentration Risk Nobody Is Talking About
The flip side of SMH's edge is concentration risk. If Nvidia disappoints on its May 27 fiscal Q1 print, SMH will move meaningfully more than SOXX. The historical beta of SMH to Nvidia is roughly 0.8 — a one-percent move in Nvidia is worth roughly 80 basis points of move in SMH. That is more leverage than most allocators realize when they buy the single ticker, and it is the primary reason a diversified allocator might choose SOXX or the iShares Semiconductor UCITS alternatives despite weaker flow momentum.
There is also a product-innovation angle. The newer premium-income semiconductor products and covered-call variants have started to attract attention among yield-focused retail traders. Those products sacrifice upside for income but deliver double-digit headline yields in a still-trending market. They are not a replacement for SMH, but they are a useful tool for traders who want partial exposure with defined downside buffer. That segment of the market is growing faster than the headline SMH numbers suggest, and it is worth watching for any allocator building semiconductor exposure for the first time.
The TSMC Hand-Off
Thursday's TSMC print was the single most important event for the semiconductor ETF complex in April. With advanced chips at 75% of total wafer revenue and Q2 revenue guided to a 10% sequential step up, the fundamental base case for SMH is intact. TSMC raising full-year 2026 revenue guidance to more than 30% growth confirms that the capacity bottleneck is matched by actual pricing power. Gross margin of 66.2% — above the high end of the guide — says TSMC is not absorbing more Arizona drag than it can price through.
The 3% selloff in TSM on Friday is worth interpreting. Some of that was profit-taking on a well-telegraphed print. Some was concern about the manufacturing expansion pace, particularly in the U.S. Either way, SMH's other large weights — Nvidia, Broadcom, AMD, Applied Materials, ASML — largely shrugged off the TSM move. That divergence is a constructive sign for SMH as an allocation vehicle. The fund is not tracking TSM line-by-line; it is tracking the broader capex trade.
What Matters Next
The calendar is densely packed with catalysts. ASML reports on April 21 and is the single cleanest tell on leading-edge lithography demand. Intel's April 24 print will test whether the foundry-services pivot is generating revenue at scale or still consuming capex. AMD's May 6 call will frame MI300X uptake against the Nvidia Blackwell cycle. Nvidia's May 27 earnings remain the pivotal event.
For traders positioning SMH, the clean read is straightforward. A close above $295 on elevated volume extends the April rally toward the all-time high in the $310 range. A break below $275 with three consecutive negative flow days signals the thematic trade is tired and allocators should consider rotating into SOXX for diversification or into bond ETFs for dry powder. Between now and May 27, every piece of CoWoS packaging capacity commentary and every hyperscale capex update is material. Treat SMH like an option on Nvidia earnings, because that is functionally what it has become.
