“Wiiiilsooon!” That’s the only part of Cast Away I know. I’m not a Tom Hanks fan.

I am a Mike Wilson fan, though. Or at least I was. Then he stopped making mainstream financial media headlines and I promptly forgot he existed. Blame Bloomberg for not keeping him above the digital fold.

On Monday, in the post-holiday lull stateside, Wilson’s name elbowed its way into a relatively prominent slot on the financial front pages. My elder millennial brain did what any elder millennial brain would do: Conjured Hanks yelling after his volleyball friend.

As it turns out, the latest from Wilson (the equity strategist not the volleyball) was in fact worth a mention. Mike reckons we’re in the early stages of a rotation that’ll favor the hyper-scalers again versus semis, which are overdue for a correction.

The figure above gives you a sense how the rally in key SOX constituents (in some cases the moves were parabolic) drove historic outperformance versus the vaunted septet, despite the overlap between the two cohorts via the biggest company on Earth.

I’ve been over this at length, most exhaustively in “Hyper-Drag” and “Obvious Things,” which are still worth reading if you missed them. The key point is that hyper-scaler underperformance and semi mania are two sides of the same coin.

As SocGen’s Manish Kabra put it in June, “In effect, AI capex is driving a transfer of cash within the value chain from hyper-scalers funding the build-out to semis monetizing it.”

But from a rate of change perspective, hyper-scaler capex will likely peak soon, and after an inflection for the ages, semi EPS estimates and revisions breadth might’ve seen the highs too. Wilson addressed those points on Monday.

“The sharp divergence between the performance of the hyper-scalers and semis was likely unsustainable given the latter is dependent on the former,” he said. “In fact, these groups tend to ebb and flow in opposite directions near the end of these divergences, which then leads to the inevitable reconciliation as the hyper-scalers either tone down their capex guidance or indicate they might be going in another direction.”

The simple figure above shows you AI semis specifically versus the hyper-scalers since end-March.

Although Wilson by no means believes the capex cycle’s over, he sees potential for the hyper-scalers to “temper expectations on the rate of change on capex growth given the poor performance of their stocks over the past few months.”

In addition, Morgan Stanley likes the hyper-scalers because, as Wilson expounded in the same note, they have “attractive optionality within the AI ecosystem [given] strong core businesses, the ability to participate in/lead the agentic application layer, as well as an under-appreciated cost cutting lever.”

All that, and they’re among the largest and best-run companies in the entire history of capitalism. Even Oracle, which tends to get the side eye in this discussion, is a storied blue chip enterprise, after all.

Wilson pointed to Meta’s plans for selling excess compute as a harbinger. A peak in the rate of change for semi revisions breadth is “a chief reason” the sector could enter a correction, he said Monday. “Meta provided the rationale for why that might happen now.”