I’m updating my memory thesis after Jensen’s Vera CPU comments last night “Micron benefits if it wins more HBM at NVIDIA.” That’s true, but incomplete. More impt now -> MU benefits because Vera expands the TAM for high-performance DRAM beyond HBM. Jensen framed Vera as a new $200B CPU TAM. CPU platforms will need LPDDR5X and high-bandwidth server memory MU isn't the dominant HBM supplier. SK Hynix and Samsung are still the leaders MU’s strength is that it has a broad DRAM portfolio across HBM, LPDDR, DDR5, and GDDR at a time when AI is pulling capacity away from every major DRAM category. So the bull case shifts - Old framing: MU needs a huge NVIDIA HBM win. - Updated framing: MU benefits from Vera-driven LPDDR5X demand, modest HBM volume, and broader DRAM tightness as AI consumes
View on X ↗