This Popular Semiconductor ETF Has a Big Chip in Its Bull Case. How to Tell If It’s Broken Beyond Repair.

I have a chart that will knock your socks off. It’s none other than the very popular iShares Semiconductor ETF (SOXX), and I’ve marked in purple a very important point in the long up cycle. That’s a triangle, or wedge. And if SOXX breaks down through it without an instant recovery, it could open the stock market floodgates. 

For now, it is simply a marker. And it won’t take long to resolve, one way or the other. That’s the usual nature of this type of pattern, once it hits that most narrow section at the far right of the chart. 

However, investors in SOXX are not flinching. At least not yet. From May 6 of this year through Monday, SOXX is flat in price. But during that time, assets under management are up 27%. What was a $9 billion ETF about 17 months ago is now at $41 billion, after being a stone’s throw from $50 billion earlier this year. 

That in itself might be a contrarian indicator for semiconductor stocks and the AI trade they are currently driving. Or, it could be a sign that this time around, investors are true believers in the space and will provide an “AI put option” effect, stopping SOXX from losing its footing.

The semiconductor industry has clearly been the engine of global equity markets for the past several years, carrying broad indexes to repeated record highs. However, SOXX has a chart that screams “look at me now, don’t wait.” So let’s do that. Because as I analyze it, the AI trade now sits at a critical fork in the road. The narrative surrounding the chip industry is facing its most rigorous inspection in recent memory. 

We made it past earnings season. But now, the scrutiny will only increase. Market participants are increasingly asking themselves whether the narrative that drove the semiconductor boom has run out of gas, or if the current hesitation is merely a necessary pause before the next move higher.

Recent pressure on this star industry stems from shifting expectations around corporate spending and industry cyclicality. Let’s face it, the huge buildout of AI infrastructure has been funded by a handful of megacap technology giants. Institutional investors are asking harder questions about when these multibillion-dollar capital expenditures will generate tangible revenues. Frankly, it all has a very 1999 “Y2K” feel to it. 

Any hint that cloud titans are slowing their hardware procurement cycles creates immediate downward pressure across the chip supply chain. And, while advanced AI accelerators dominated headlines, broader semiconductor categories, such as automotive, industrial, consumer electronics, and traditional analog chips, are struggling with prolonged inventory issues, and sluggish end-market demand.

Furthermore, big tech’s clients are aggressively developing proprietary, custom ASICs (application-specific integrated circuits) to reduce their reliance on off-the-shelf graphics processors. As cloud providers move workloads onto custom chips, pricing power and gross margins across broad semiconductor indexes face long-term erosion.

Still, we didn’t get this far to simply dismiss as “overvalued” what is a remarkable, forward-looking project. Semiconductor stocks are certainly cyclical, but cycles can last much longer than bears would hope.

To the point, the transition from early AI model training to widespread enterprise inference is going to demand exponentially more computing power. “We need more compute” is a phrase I hear every business day in the media. And if this is to be the next big industrial revolution, they’re right. We are forever seeking more of what chip producers do.

As AI agents, video generation, and autonomous robots scale globally, the physical demand for advanced silicon, high-bandwidth memory, and complex packaging continues to expand. Building cutting-edge chip fabrication facilities requires hundreds of billions of dollars and years of specialized engineering. The few companies that control the advanced node ecosystem possess deep economic moats, unmatched pricing power, and essential strategic importance to national infrastructure.

Beyond enterprise computing, long-term trends toward vehicle electrification, defense modernization, and smart industrial automation guarantee that the content value of semiconductors per device will continue climbing over the next decade. But while there’s strong rationale for both the bull and bear cases, my chartist tendencies force me to wave a yellow caution flag. 

SOXX is coming off a three-year period in which it gained 300%. The only other time this century where we’ve seen a move of 250% or more in such a short period of time was the period ending at the very beginning of 2022. The next move of consequence was a 44% drop in under 10 months’ time. SOXX did not recover from that early 2022 peak until April 2025. 

Translation: Even the start of the AI-inspired SOXX move this time around was really just getting long-term semiconductor investors back to break-even. 

However, what I’m watching just as carefully here is what we see immediately above. I am regularly analyzing the technical behavior of SOXX and other semiconductor ETFs. 

But just as important to me is whether or not the top stocks, shown here, are moving in sync. If so, that’s a much bigger indication of a market top, not only one in SOXX or even the Invesco QQQ ETF (QQQ). That’s more than 60% of SOXX’s total assets up there, across only 10 stocks. If you’re looking for where to spot the market’s inertia, or lack thereof, heading into the fall, I think that list above is a great place to start. 

Rob Isbitts is a semi-retired CIO, former fiduciary investment advisor, and Barchart columnist. Check out his other work at ETFYourself.com (featuring the Fresh Charts weekly trading post), and ROAR.PiTrade.com, helping investors to better-manage their own portfolios.