By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
The panel is largely bearish on SMH, citing high concentration risk (especially in Nvidia), rich valuation (26x forward P/E), and potential risks to growth such as inventory gluts, export curbs, and AI capex normalization. However, there's debate on whether SMH's earnings can hold up better than broad semis cycles imply, given high-margin AI accelerators and potential capacity tightness.
Risk: Concentration risk, particularly the high exposure to Nvidia (18%) and the top six holdings (55%), makes SMH vulnerable to underperformance if these names stumble.
Opportunity: If AI capex demand remains strong and capacity tightness persists, SMH's high-margin AI accelerators could continue to drive earnings growth, supporting the fund's multiple.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Unless you've been completely off the grid for the past few years, you probably know that semiconductor stocks have been some of Wall Street's hottest investments.
Over just the past three years, the VanEck Semiconductor ETF (NASDAQ: SMH) has returned an astonishing 327%. That far outpaces the 117% return of the Invesco QQQ ETF (NASDAQ: QQQ) and the 85% return of the Vanguard S&P 500 ETF (NYSEMKT: VOO).
| Continue » |
Now, semiconductors are facing competing forces. On one side, the tailwind of the artificial intelligence (AI) buildout is still very much in play. On the flip side, high inflation, slowing GDP growth and a Fed that may soon need to hike rates adds several complications to the outlook. Plus, the portfolio trades at 26 times next 12 month earnings estimates.
Should investors keep buying the chip demand story or start dialing back their expectations?
The demand for semiconductor chips isn't in question. Pretty much every major name in the sector is struggling to keep up with demand. The big question is whether shares are fairly priced based on growth conditions, or are they overdue for a breather.
The global semiconductor industry is forecast to reach $975 billion in sales in 2026, a 26% increase from 2025. Sales could hit the $2 trillion mark within the next decade.
Semiconductor manufacturers are going to need to spend substantial amounts of money on their own AI infrastructure, but the return on investment should be strong for the foreseeable future. Given that so many companies are still in the very early stages of their own buildouts, there's little question that chip demand will remain durable for years, if not decades.
There are a couple of things to consider before jumping into the VanEck Semiconductor ETF.
First, the fund is market-cap-weighted, which means you'll be buying a very top-heavy portfolio. Nvidia alone is nearly 18% of the portfolio. The top six holdings, which also include Taiwan Semiconductor Manufacturing, Broadcom, Intel, Advanced Micro Devices, and Micron Technology, account for 55%.
This is a big mega-cap play based on the fortunes of just a few companies.
Second, the fund's valuation may or may not be a concern. If earnings growth rates can continue at the rates they have been in this sector, I doubt many investors will worry about paying a 26 P/E multiple. But when we get to that point where growth rates start slowing, and we're past the peak acceleration period, it won't be surprising to see valuations potentially shrink quickly.
Four leading AI models discuss this article
"Extreme concentration and stretched valuations leave SMH vulnerable to rapid de-rating if AI spending growth slows."
SMH's 327% three-year outperformance stems from AI-driven chip demand, yet the 26x forward P/E offers scant margin if growth moderates. Market-cap weighting leaves the ETF exposed, with Nvidia at 18% and the top six names exceeding 55% of assets. Macro complications such as potential Fed rate hikes, inflation, and slowing GDP could pressure capex budgets sooner than expected. Long-term sales forecasts to $975 billion by 2026 ignore near-term risks like inventory digestion or export curbs on advanced nodes. A valuation reset appears likely once peak AI acceleration passes.
Sustained AI infrastructure buildouts could deliver earnings growth above 25% for several more quarters, allowing the multiple to hold or expand even amid higher rates.
"SMH is pricing in 5+ years of peak-cycle earnings growth at current multiples, but semiconductor cycles compress faster than most investors expect once capex saturation hits."
SMH's 327% three-year return masks a valuation trap. At 26x forward P/E against a sector growing ~10-12% annually (not the 26% headline growth, which is cyclical capex-driven, not earnings), you're paying ~2.2x the growth rate—well above historical semiconductor multiples of 1.2-1.5x. The Nvidia concentration (18%) means you're buying one stock's execution risk wrapped in ETF diversification clothing. The article conflates durable *demand* (true) with durable *margin expansion* (false). When AI capex normalizes—likely 18-24 months out—multiples compress faster than earnings grow.
If semiconductor supply constraints persist longer than expected and AI capex accelerates beyond consensus (cloud capex hit $60B+ in Q1 2024), earnings growth could sustain 18-20% CAGR through 2026, justifying current multiples or even re-rating higher.
"SMH's extreme concentration in a handful of mega-caps makes it a high-beta volatility play rather than a foundational semiconductor sector investment."
The article focuses on the 26x forward P/E, but this ignores the extreme concentration risk inherent in SMH. With Nvidia representing nearly 20% of the fund, SMH is essentially a leveraged bet on AI infrastructure spending rather than a diversified semiconductor play. While the 2026 sales forecasts are impressive, they assume no cyclical correction in legacy chip markets like automotive or industrial, which are currently showing inventory gluts. Investors are paying a premium for growth that is already priced in; if Nvidia misses even one quarterly guidance beat, the index-level volatility will be disproportionate given the lack of diversification in the top six holdings.
If we are in the early innings of an industrial AI revolution, the current 26x forward P/E is actually a bargain compared to the 50x-plus multiples seen during the dot-com bubble, suggesting significant room for further multiple expansion.
"SMH's upside hinges on a persistent AI-capex boom and ongoing leadership from Nvidia/TSMC; any turn in that trend or macro shock could trigger fast downside via concentration and multiple compression."
SMH's 3-year run (327%) is impressive, but the exposure is highly concentrated: Nvidia ~18% and the top six holdings account for 55%, so a few names largely drive returns and risk. The 26x forward P/E looks rich if growth slows or AI capex moderates. Macro headwinds—sticky inflation, potential rate hikes, and geopolitics around Taiwan—add downside risk. The article omits concentration risk and the risk of multiple compression if earnings growth decelerates; a more diversified or risk-managed semis exposure could be prudent.
Against the bullish take: AI-driven growth may cool or capex could moderate, triggering multiple contraction. SMH's top-heavy exposure to Nvidia/TSMC means a setback in those names could depress the entire ETF more than broader tech.
"Bifurcated margins between AI and legacy chips could support SMH valuation longer than assumed amid TSMC allocation shifts."
Claude's point on cyclical capex-driven growth ignores the structural shift where AI accelerators command 60-70% gross margins, far above the 40% for mature nodes. Combined with Gemini's inventory glut warning in autos, this suggests SMH earnings could hold up better than broad semis cycles imply, but only if TSMC's capacity allocation prioritizes high-end products through 2025. The 26x multiple might then prove sustainable rather than a trap.
"AI accelerator margins are not structurally insulated from cyclical capex normalization; they're just more volatile."
Grok's 60-70% gross margin claim for AI accelerators needs scrutiny. TSMC's Q1 2024 gross margin was 54%, not materially higher than historical levels. The real risk: if AI capex normalizes faster than capacity reallocation, high-end utilization drops sharply, and those margins compress harder than legacy nodes because they're already priced for perfection. Nobody's modeled a 12-month capex pause scenario.
"The sustainability of SMH's valuation depends on designer margins, which face imminent erosion from hyperscaler custom silicon adoption."
Claude, your focus on TSMC’s 54% gross margin misses the critical distinction between foundry margins and fabless designer margins. Nvidia’s gross margins are currently hovering near 75-78%, not the foundry margins you’re citing. The risk isn't just utilization; it’s the sustainability of those designer margins as custom silicon from hyperscalers like Google and Amazon begins to displace merchant silicon. If custom AI chips commoditize the accelerator market, the entire valuation thesis for SMH’s top-heavy holdings collapses.
"Scarcity-driven demand and moat effects could sustain high valuations for SMH even if near-term AI capex growth moderates."
Claude, you assume AI capex normalizes quickly and multiples compress as a result. I’d push back: if capacity tightness persists into 2025, pricing power for high-end AI accelerators could stay durable, keeping margins and earnings growth above broad semis. That scarcity effect could support, or at least cushion, SMH’s multiple, even with Nvidia-heavy risk. Concentration risk remains, but the thesis isn’t dead just because the cycle moderates.
The panel is largely bearish on SMH, citing high concentration risk (especially in Nvidia), rich valuation (26x forward P/E), and potential risks to growth such as inventory gluts, export curbs, and AI capex normalization. However, there's debate on whether SMH's earnings can hold up better than broad semis cycles imply, given high-margin AI accelerators and potential capacity tightness.
If AI capex demand remains strong and capacity tightness persists, SMH's high-margin AI accelerators could continue to drive earnings growth, supporting the fund's multiple.
Concentration risk, particularly the high exposure to Nvidia (18%) and the top six holdings (55%), makes SMH vulnerable to underperformance if these names stumble.