The rush into artificial intelligence has sent a lot of investors hunting for the next bubble, and semiconductor stocks are now in the crosshairs. But the VanEck Semiconductor ETF, known by the ticker SMH, still looks more like a broad bet on chip demand than a one-way wager on AI fever.
SMH is entirely invested in semiconductor companies and holds about 25 of them, including Nvidia and Taiwan Semiconductor Manufacturing. Nvidia recently traded at a forward-looking price-to-earnings ratio of 24, while Taiwan Semiconductor Manufacturing was at 26, levels that are not far removed from the broader market and a long way from the kind of valuation excess that has historically marked a peak.
That matters now because AI spending has become one of the market's biggest stories, with some investors worried the buildout could unwind and drag semiconductor names down with it. Yet the ETF's 15-year average annual gain of 25.8% came through multiple market cycles, and the stock market itself advanced by double-digits from 2019 to 2025 and rose by nearly 18% last year. That backdrop has left little room for easy calls about where the next crack will appear.
The comparison that keeps surfacing is Microsoft in 1999, when its price-to-earnings ratio peaked at 66 before the 2000 crash. Microsoft's more recent ratio was 26, and the larger point is that not every expensive-sounding technology bet is priced for disaster. Microsoft and Meta Platforms are described as not having steep valuations, and the article says hyperscalers' heavy investment in AI infrastructure is generally being funded from earnings rather than borrowings.
The tension for SMH is that the ETF is not all-in on AI. Its holdings also lean on chips for gaming, automobiles, computers and other uses, which means the fund is tied to a wider electronics economy, not just the fate of data-center demand. That makes many of its holdings seem more undervalued than overvalued, even as the AI trade dominates the conversation. For investors trying to judge smh stock today, the bigger question is not whether AI has momentum — it does — but whether the chip industry is already being priced as if all of that momentum ends badly.
