At first glance, the 2026 divergence between the market fortunes of chip designer Nvidia Corp (NASDAQ:NVDA) and Salesforce Inc (NYSE:CRM) looks like a simple stock story.
Nvidia’s stock is up about 13% this year but relative to Salesforce, its outperformance almost jumps five-fold to 64%. That widening gap, however, isn't just about Nvidia and Salesforce. Instead, it's a reflection of a much bigger and deeper fracture that’s dividing the tech sector ― the semiconductors versus software divide.
As Bloomberg recently reported, one of the most successful trades in a choppy year for tech has been to buy semiconductor stocks and sell software shares, with the gap between the two only widening in 2026. What looks like a stock-specific story is, in reality, a structural divide.
The Real Trade Is Playing Out In ETFs
Zoom out, and the same divergence is also clear at the ETF level. The VanEck Semiconductor ETF (NASDAQ:SMH), a key proxy for chip stocks, has gained around 35% year-to-date. But when compared with the iShares Expanded Tech Software ETF (BATS:IGV), its relative outperformance stretches to roughly 68%.
Bloomberg said that the performance gap between chipmakers and software companies has reached record levels this year, both in absolute terms and on a relative basis. Analysts are saying that “growth and margins for semis keep getting re-priced higher and growth and margins for software keep getting re-priced lower.”
That framing is critical. Investors aren't just making money by picking winners, they're increasingly profiting from the widening gap between two segments of tech that are moving in fundamentally different directions. ETFs, in turn, have become the most efficient way to express that view, allowing for broad exposure to each side of the trade without relying on individual stock calls.
The divergence is being powered by artificial intelligence (AI).The semiconductor companies are expected to deliver significantly stronger earnings growth in the coming years, driven by relentless demand for AI chips and data center infrastructure.
A Powerful Trade—Expressed Through ETFs
The Nvidia-versus-Salesforce gap makes for a compelling trade opportunity, but it is ultimately just a snapshot of a larger market dynamic. The real trade lies in the growing divide between semiconductors and software, and in how investors are using ETFs like the SMH and IGV to capture it.
That's especially relevant in a market where dispersion within tech is high. A single earnings miss or beat can sharply move individual stocks, but ETFs smooth out that idiosyncratic risk while still reflecting the broader trend. This has made long semiconductors vs. short (or underweight) software via ETFs one of the most efficient ways to play the AI divide.
Crowding Risk—And What ETF Signals Are Saying
There is growing skepticism about how long this divergence can last. The semiconductor rally has pushed valuations higher and driven the group into what some strategists describe as "extreme" territory, raising the risk of a reversal. At the same time, some investors argue that the selloff in software may be overdone, leaving room for a rebound if sentiment shifts.
Semiconductor ETFs have seen strong inflows alongside their rally, a sign that investors continue to pile into the AI infrastructure theme. At the same time, software ETFs have lagged, with weaker price action suggesting cautious sentiment. This growing imbalance is precisely what has fueled the trade—but it also increases the risk of a sharp unwind if positioning becomes too crowded.
For ETF investors, that makes flow data, relative performance, and momentum trends critical signals to watch. A slowdown in inflows into semiconductor funds or stabilization in software ETFs like IGV could indicate that the gap is beginning to narrow.
For now, however, the trend remains firmly in place. In an AI-driven market, capital continues to flow toward the companies building the infrastructure. And for investors, the most effective way to capture that shift may not be choosing individual winners, but trading the divide itself.
Image: Shutterstock
Login to comment