What If Chip Stocks Aren’t in a Supercycle After All?

Semiconductor stocks have looked virtually unstoppable this year, and while the year-to-date rally looks unsustainable and adds more air into a bubble that was already on the verge of becoming far larger than in any other corners of the AI trade, questions linger as to what could prick the bubble.

That is, if it is a bubble at all. Cycles are hard to time. No doubt about that — especially as we enter a pivotal point in an AI boom or fourth industrial revolution that could present new opportunities, but, at the same time, elevated risks and considerable backlash, given the disruptive impact on jobs and resistance facing data center builds.

Whether you’re in the camp that thinks semis are in a cycle, a longer-lived supercycle, or a paradigm shift that entails changing how we look at the semis, given the sustained, structural shift in demand for chips powering the next phase of the AI revolution, there’s no shortage of back and forth when it comes to the state of the semi market as share prices begin to swell while valuation multiples look, believe or not, relatively cheap.

Cycles are scary, especially when it comes to red-hot names

The history of cycles tells us that it’s a bad idea to buy such stocks when big gains are made, and multiples look cheaper. But, in my view, the price-to-sales (P/S) multiple suggests semi stocks aren’t a “steal,” but are going for a mildly-heated premium. Of course, the hot topic has to be whether semis are in a cycle. It all comes down to whether the AI buildout continues to accelerate and if the year-over-year efficiency gains can still impress.

There’s no question that there’s a healthy dose of skepticism clashing with the euphoria surrounding corners of the AI trade. High CapEx is haunting some stocks to this day — look at hyperscalers like Microsoft (NASDAQ:MSFT | MSFT Price Prediction).

At the same time, many investors are now calling for ROIs and disciplined growth. And one has to think that companies are going to start listening as they cut costs, both to shore up capital to spend on AI and because AI is improving productivity in a way such that a smaller headcount is necessary. In any case, I think the scariest scenario to envision for semi investors is one where the Mag Seven start spending less and not more in 2027 and beyond.

What’s next for the semis?

At this juncture, it seems like the CapEx figures can only go higher, perhaps much higher. But what if higher interest rates from the new Fed chair come into play? Could higher rates cause a shift in the aggression of AI strategies? And what would the impact be on the semis?

It’s not quite clear. A digestion phase could precede as investors await the big monetization upswing. And it’s not quite clear what such a digestion would entail for the state of the AI trade. If CapEx does come down, maybe semis could take a hit while the hyperscalers, which have arguably already served their time, could hold their own.

Either way, though, I wouldn’t view that as the end of the cycle for semis, but as a vicious breather — one that might open up a window to buy before the next phase. Indeed, fear could get out of hand if such a rotation were to happen. So, a strong stomach is needed, especially if the “bullwhip” effect plays out, which sees growth slow suddenly. Of course, I could be wrong, and the supercycle could last into the 2030s. But, in my view, I think it’s wise to be aware of the risks that come with the potential rewards from the space.

In short, I think the cycle is alive and well, but perhaps a bit overdue for a correction at some point. Over the long run, though, a structured secular shift is also more than possible as “silicon as infrastructure” continues to be the winning way to bet on AI.