LG Doucet speaks with Milk Road analyst Melvin about his largest AI infrastructure positions: Nebius, Micron, Nvidia, and Oracle. Melvin explains how demand for compute and high-bandwidth memory supports his thesis, while Nvidia’s broader data center stack and Oracle’s cloud backlog offer exposure to expanding AI usage.
The discussion also covers the risks behind these positions, including data center construction delays, power constraints, heavy capital requirements, and an eventual increase in memory supply. Melvin explains why he takes profits in volatile names, buys pullbacks, and is gradually shifting attention from well-known bottleneck trades toward hyperscalers that could begin showing more AI-driven revenue.