Thoughts on the Market; Networking Survey; Joe’s 2Q HW Previews and Pre-EPS Company Call Takeaways

Research evidence excerpt

Thoughts on the Market; Networking Survey; Joe’s 2Q HW Previews and Pre-EPS Company Call Takeaways

and the rotation we’re seeing back into laggard Mag 7 includes: 1) token costs/frontier model

viability/open source debate; token costs per task are coming down as model choice/efficiency becomes a focus, something that

comes up in every conversation (and risks overpowering the market today and next week as generalists dissect the launch of

Kimi-K3). Will frontier model revenues continue surprising on the upside (we think so), and if not, who will pay for all these

chips? “Jevon’s Paradox” is losing its power as an incantation; 2) asymmetery; folks we talk to pretty universally expect this to

be a super quarter for hardware & semis prints, so it may be tough to surprise positively. How much optimism will companies

betray on their 27 outlooks – particularly in the face of all kinds of deployment bottlenecks and shortages? Before this pullback, at

least, downside risks didn't seem well-discounted. The tepid market responses to ASML & TSM results which specs generally

thought were quite strong double-underlines this; 3) Heavey in trading has talked about simple lack of technical support/high

VARs, and mid-year positioning shifts; even yesterday, quite a few names had downside even to their 50MA (underscoring how

so many names have gone parabolic), with incremental buyers scarce – though unfortunately the supply they’ve been seeing in

the AI complex has more been out of Asia, and they’re not seeing major US pukes yet. The trading desk continues to see a nice

bid in hyperscalers that doesn’t make a ton of sense to me, but I guess VOL is lower here, and they’re in some ways the inverse of

the long memory trade.