Let’s say year-to-date rally in semis already priced a higher share of memory in hyperscaler capex for 2027. There are two paths from here: 1- $NVDA and other designers will have to increase prices to keep their margins, which will require upside revisions to hyperscaler capex estimates post-2028 if volumes won’t also be sacrificed. I think aggressive upside revisions are unlikely as their free cash flows are almost drained, they won’t be too aggressive with debt or dilution. 2- Designers eat some of the price increases in memory, which means declining margins, which means multiple compression and likely end of the semiconductor super cycle. I am not a bear, but the chances that semi super-cycle will make another leap upward leap from here is low. Time to be more cautious than aggres
View on X ↗