Wedbush's Dan Ives — one of tech's most visible bull-side analysts — is characterizing AI memory chips as the 'Golden Child' of the AI infrastructure cycle, citing a demand-to-supply imbalance he quantifies at 15-to-1. The primary beneficiary he's pointing to is the high-bandwidth memory (HBM) space, where Micron Technology (MU) and its peers are racing to expand capacity to meet insatiable demand from hyperscaler AI training and inference workloads.

MU's most recent fiscal year (ending August 2025) shows revenue of $37.4B, up 48.9% year-over-year, with gross margins at 39.8% and diluted EPS of $7.59 — numbers that validate the demand narrative on their face. KLA Corporation (KLAC), a key equipment supplier enabling memory chip manufacturing yield and process control, also posted strong results with $12.2B in revenue (+23.9% YoY) and a 33.4% net margin, reflecting the upstream capex wave.

The bull case centers on the structural argument: if AI infrastructure spending continues to compound and HBM supply cannot ramp fast enough, memory pricing power and margins should hold or expand, driving further earnings beats for MU. The bear case, however, is that Ives is a famously optimistic analyst, and the '15-to-1' figure is an anecdotal claim without disclosed methodology — memory has a well-documented history of violent cyclical reversals when supply eventually catches up.

What to watch: MU's next earnings print for HBM revenue mix and forward guidance commentary on supply additions; any signals from Samsung or SK Hynix on capacity ramp timelines; and broader AI capex signals from hyperscalers like Microsoft, Google, and Amazon, which are the ultimate demand anchors for this thesis.