Chip stocks were a safe AI play. Now they’ve turned into the market’s pain trade.

Semiconductor stocks have long been treated as a relatively straightforward way to gain exposure to artificial intelligence, on the logic that chipmakers supply the hardware regardless of which software companies ultimately win. That positioning is now being tested, as the same crowded trade that drew investors in has become a source of market pressure.

An important nuance is that a slower pace of AI model development would not automatically translate into weaker chip demand. Spending on data-center capacity is driven by multiyear infrastructure plans, and orders are often committed well in advance. A deceleration in headline model releases, in other words, is not the same as a slowdown in capital expenditure.

What matters more for investors is the set of risks sitting alongside that demand: stretched valuations after a strong run, concentration in a handful of large names, and the possibility that expectations have outrun what earnings can confirm. For now, the sector’s direction is likely to hinge less on AI headlines and more on whether reported capital spending and guidance continue to justify current pricing.