July 2026 · Interim Note
One Theme, Three Different Reactions
Last week's interim note was about a record Dow alongside a rough two days for my semiconductor sleeve. This one is about something more specific: the same underlying AI-infrastructure story showing up across three of my holdings, and moving each one differently.
Three data points, one theme
First, Alphabet joined the Dow Jones Industrial Average this week, replacing Verizon — the fifth "Magnificent Seven" member now inside the index. Buried in the same coverage: Google reportedly told Meta back in March it couldn't sell Meta as much Gemini AI capacity as Meta wanted, simply because Google didn't have the room. Second, Meta's own stock surged 9% on the "Meta Compute" cloud-business report, then gave back roughly half of that the very next day — a sharper, faster version of the reaction I already wrote up in NBIS and CRWV when the same story broke. I wrote both up in full on the News page: the Alphabet/Dow piece and Meta's own two-day round trip.
Why three different reactions from one theme is the actual lesson
GOOGL, META, NBIS, and CRWV all touch the same AI-infrastructure story this month, and none of them moved the same way: NBIS and CRWV fell hard as suppliers threatened by a customer-turned-competitor; META round-tripped nearly 10% in two days as the market argued with itself about whether the same move helps or hurts Meta; GOOGL barely reacted to its own capacity-constraint detail at all, arguably the most fundamentally interesting fact of the three. A single macro theme does not move a portfolio uniformly — which is both the argument for diversification within a theme and the reason "thematic clustering" is still a real risk even when individual reactions look uncorrelated day to day. The July letter flagged clustering as a structural risk; this week is a reminder that it doesn't always look like everything falling together on the same day — sometimes it looks like the same story pulling four positions in four different directions, which is easy to mistake for diversification when it isn't.
No position changes from this. The next full letter will report whether any of this — the semiconductor pullback, Meta's cloud pivot, or Alphabet's capacity constraints — turned into something that actually changes a thesis, rather than just adding texture to one already on the record.