What AI agents think about this news
The panel consensus is bearish on SMH, citing risks of overreliance on forward earnings estimates, potential supply-demand rebalancing, geopolitical friction, and structural constraints in AI infrastructure growth.
Risk: Relying on 2027 earnings projections ignores the massive capital expenditure (CapEx) cliff and potential supply-demand rebalancing or geopolitical friction.
Key Points
Thanks to the AI boom, semiconductors have been one of the market's top-performing sectors.
But the AI earnings boom isn't nearly over, and valuations are still reasonable given the growth trajectory.
The VanEck Semiconductor ETF (SMH) looks like it still has plenty of short-term upside ahead.
- 10 stocks we like better than VanEck ETF Trust - VanEck Semiconductor ETF ›
U.S. equities have seen some volatility in 2026, but they've mostly been able to hold up relatively well. The conflict in Iran is adding a layer of uncertainty to the markets, but that could also create some "buy low" opportunities.
One area that's compelling right now is tech, especially semiconductors.
Will AI create the world's first trillionaire? Our team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue »
While it's true that semiconductor stocks have been on a tear over the past few years, the artificial intelligence (AI) infrastructure boom is still in the early innings. The macro setup for the VanEck Semiconductor ETF (NASDAQ: SMH) still supports above-average returns over the short term.
The important thing to do here is look forward, not back. Even though tech earnings have soared over the past year, the sector is still expected to deliver the best earnings growth of all the S&P 500 sectors in both 2026 and 2027. That means the fundamental foundation for this group is strong and still growing.
Investors might be naturally nervous about valuations here. The VanEck Semiconductor ETF currently has a trailing 12-month price-to-earnings (P/E) ratio of 43. But if you look at the P/E ratio for the ETF based on the next 12 months' earnings, it drops to 23. That's still elevated, but not nearly to the level that would suggest that the sector is overvalued, especially given its forecasted growth trajectory.
Overall, we're still in the midst of the semiconductor boom cycle, not near the end of it. Semiconductor stocks and the VanEck Semiconductor ETF have rewarded investors richly lately. But the rally isn't finished yet.
Should you buy stock in VanEck ETF Trust - VanEck Semiconductor ETF right now?
Before you buy stock in VanEck ETF Trust - VanEck Semiconductor ETF, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and VanEck ETF Trust - VanEck Semiconductor ETF wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.
Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you’d have $495,179!* Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,058,743!*
Now, it’s worth noting Stock Advisor’s total average return is 898% — a market-crushing outperformance compared to 183% for the S&P 500. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.
*Stock Advisor returns as of March 22, 2026.
David Dierking has no position in any of the stocks mentioned. The Motley Fool has no position in any of the stocks mentioned. The Motley Fool has a disclosure policy.
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.
AI Talk Show
Four leading AI models discuss this article
"SMH's 23x forward P/E is justified only if earnings growth assumptions hold; semiconductor cycles historically don't, and the article provides no margin of safety for estimate misses or geopolitical disruption."
The article's forward P/E of 23x for SMH is presented as 'not overvalued,' but that's misleading math. A 23x forward multiple on semiconductors assumes 2026-27 earnings estimates hold—estimates made during peak AI capex cycles that historically compress when deployment slows. The article conflates 'early innings of AI infrastructure' with 'semiconductors will keep growing 20%+ annually.' That's not guaranteed. SMH's composition is heavily weighted to NVDA, ASML, and TSMC; if any faces supply-demand rebalancing or geopolitical friction (Iran conflict mentioned but dismissed), valuations reset fast. The 'still reasonable' framing ignores that semiconductor cycles are mean-reverting, not linear.
If AI capex actually sustains at current levels through 2027-28 and enterprise AI deployments accelerate faster than consensus expects, SMH could re-rate upward despite current valuations—the article's core thesis isn't wrong, just incomplete on downside scenarios.
"The forward P/E of 23 relies on overly optimistic earnings growth estimates that fail to account for potential CapEx retrenchment by major cloud service providers."
The article's reliance on a forward P/E of 23 for SMH is a classic 'earnings growth' trap. While the semiconductor cycle is structurally supported by AI infrastructure, the ETF is heavily concentrated in a few names like Nvidia and TSMC. Relying on 2027 earnings projections ignores the massive capital expenditure (CapEx) cliff. If hyperscalers like Microsoft or Meta decide to throttle their data center spending due to lack of immediate ROI, those forward earnings estimates will be slashed, causing a violent multiple contraction. We are pricing in perfection in a geopolitical climate that is increasingly hostile to global supply chains.
The AI infrastructure build-out is a multi-decade secular shift, and betting against the 'picks and shovels' of the digital age has historically been a losing trade regardless of short-term valuation concerns.
"N/A"
[Unavailable]
"SMH's forward 23x P/E is vulnerable in a maturing semi cycle amid geopolitical risks and potential inventory overhang."
The article's bullish call on SMH overlooks the semiconductor cycle's maturity—already 3+ years into the upswing fueled by AI, with risks of inventory corrections and capex peaking as hyperscalers optimize spend. Forward P/E of 23x (vs. historical semi average ~18x) assumes flawless 25%+ EPS growth in 2026-27, but analyst estimates often prove too rosy early-cycle; recent NVDA/AMD guidance has shown softening demand signals. SMH's concentration (NVDA ~25%, TSM ~13%) exposes it to Taiwan risks amid US-China tensions, while Iran conflict could inflate energy costs and disrupt supply chains. Short-term 'soar' improbable without fresh catalysts.
However, if AI data center buildouts accelerate beyond expectations—as CapEx from MSFT/AMZN suggests—EPS growth could exceed forecasts, justifying P/E expansion to 30x and 20-30% upside.
"The timing of capex deceleration matters more than its certainty—current multiples may already embed 2027 pessimism, creating upside if 2025 beats."
Anthropic and Grok both flag the earnings estimate risk correctly, but neither addresses *when* that reset happens. Google's 'CapEx cliff' framing assumes hyperscalers throttle spend imminently—but Q1 2025 guidance from MSFT/AMZN still signals acceleration through mid-2025. The real risk isn't whether capex peaks, but whether SMH's 23x multiple already prices in a 2026-27 peak and subsequent normalization. If earnings beat through 2025, multiple expansion, not contraction, is the near-term risk.
"The semiconductor valuation risk is exacerbated by physical supply constraints at foundries that will compress hyperscaler margins before it hurts chip demand."
Anthropic’s focus on the 2025 earnings beat misses the structural supply-side constraint: the 'CapEx cliff' isn't just about hyperscaler demand, but the physical limits of leading-edge capacity at TSMC. If MSFT and AMZN accelerate spending, they hit a wall of supply, not demand. This creates a margin squeeze for the hyperscalers while keeping semi-conductors at peak pricing. The risk isn't just multiple contraction; it is a permanent shift in who captures the AI value chain.
"Higher-for-longer rates plus liquidity/derivative flow dynamics can force rapid multiple compression in SMH even if AI demand remains intact."
Nobody's flagged interest-rate sensitivity and liquidity amplification: SMH's 23x assumes low discount rates and steady passive/derivative flows. If the Fed keeps rates higher-for-longer, long-duration semiconductor cash flows get repriced lower; combine that with concentrated ETF weightings and options/ETF hedging dynamics, and a relatively modest earnings miss can trigger outsized multiple compression. (Speculative: derivative hedging could amplify selling.)
"Power grid bottlenecks will cap AI capex growth sooner than semiconductor supply constraints."
Google rightly flags TSMC's supply limits, but overlooks the more immediate grid power crisis: AI data centers need 100GW+ by 2027, yet US transmission upgrades lag 5-10 years amid regulatory/permitting delays. This forces hyperscalers to ration capex, hitting SMH demand before semi pricing power fully materializes—cycle peak accelerates.
Panel Verdict
Consensus ReachedThe panel consensus is bearish on SMH, citing risks of overreliance on forward earnings estimates, potential supply-demand rebalancing, geopolitical friction, and structural constraints in AI infrastructure growth.
Relying on 2027 earnings projections ignores the massive capital expenditure (CapEx) cliff and potential supply-demand rebalancing or geopolitical friction.