Opinion · August 2026
Rotation Isn’t a Thesis, and Neither Is the Absence of One
Monday's tape had a clean, named cause underneath it: a rotation out of semiconductors and into software, flagged in real time by The Motley Fool two days before it showed up in my own book as Arm, Intel, and AMD falling while Palo Alto Networks, CrowdStrike, and Datadog rose. That's a legible story. What I want to push back on is the instinct to treat "legible" as the same thing as "meaningful" — and its mirror image, the instinct to treat an unexplained move like Datadog's as automatically less real than an explained one.
A named rotation and an unnamed round trip
The rotation itself is real and dated: chip stocks have run hard in 2026 and gotten expensive by several accounts, while software names that were sold off earlier in the year on AI-disruption fears looked comparatively cheap. Money moved accordingly on Monday. But a sector-level flow of capital chasing relative valuation isn't a verdict on any single company's fundamentals — Arm falling 5.21% on a rotation day says nothing new about Arm's licensing business, any more than Palo Alto Networks rising 5.82% validates its 96x forward-earnings multiple that the same reporting flagged as stretched just one day earlier.
Datadog is the harder case, because I genuinely don't have a story for it. It fell 19.03% on August 6 for reasons I couldn't source then and still can't now. It rose 11.48% on Monday, clawing back roughly half of that move, also for reasons I can't source. The honest read is that both numbers are just as "real" as Monday's named rotation — they moved the price I'd transact at — while carrying zero information about the business. Treating an unexplained 19% drop as more alarming than an explained 5% sector rotation, simply because the rotation comes with a headline attached, is a bias worth naming and resisting.
Where this connects to the capex debate
This rotation sits next to a question this site has spent real time on: whether AI infrastructure spending is still generating a return that justifies its size. The AI Capex Reality Check found Alphabet and Meta both spending faster than their free cash flow covers. If chip names are partly getting sold because their run has outpaced that same spending story, and software names are partly getting bought because they were oversold on a disruption fear that hasn't fully played out, Monday isn't noise so much as the market groping toward the same question from the opposite direction — without yet answering it.