Chip stocks shift from safe AI play to the market’s pain trade

For much of the AI boom, the playbook was simple: buy chips, make money. Semiconductors were the shovels in the AI gold rush, and investors treated them accordingly. That trade has now flipped on its head.

The unraveling

The cracks started showing in June, when the Nasdaq dropped more than 4% in a single session. That day was particularly unkind to semiconductor names. Micron, Intel, Super Micro, and Sandisk each cratered more than 11%.

July made June look like a warmup. The Philadelphia Semiconductor Index logged its worst week in over a year, and the broader semiconductor-versus-software ratio posted one of its ugliest monthly performances in over two decades.

Then came September 14, when a group of prominent AI industry leaders publicly called for a slowdown in AI model development, citing safety risks. The market’s response was swift and predictable: Intel, Micron, and AMD all dropped between 5% and 8% on the day.

Strong earnings, weak sentiment

What makes this sell-off particularly disorienting is that it hasn’t been driven by collapsing fundamentals. Micron, for instance, continued to post exceptional earnings growth through mid-2026. Demand for high-bandwidth memory and AI accelerators hasn’t evaporated.

The problem is valuation. Chip stocks had been priced for perfection, and when investors started questioning whether hyperscaler capital expenditures on AI infrastructure would continue at the same breakneck pace, the premium embedded in semiconductor names started to look more like a liability than a cushion.

Rising interest rates added another layer of pressure. Higher rates make future earnings less valuable in today’s dollars, which is particularly painful for growth stocks trading at elevated multiples.

August’s dead cat bounce, or real recovery?

After the bloodbath of June and July, August delivered a sharp snapback. Semiconductor stocks posted their best August in over 20 years, a recovery that briefly restored confidence in the AI hardware thesis.

The September sell-off that followed suggests the August rally was more of a pause than a pivot.

The rotation question

The nearly 20% collapse in the SOXX/IGV ratio during July was one of the clearest signals yet that the market is reconsidering whether hardware deserves its premium over software. If hyperscaler spending does moderate, or if AI safety concerns lead to regulatory constraints on development pace, software companies with recurring revenue models could offer a more stable way to play the AI theme.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.