Overview
- Micron has started shipping HBM4, a next‑generation high‑bandwidth memory that doubles bandwidth over prior chips while cutting energy use and is reported to be used in Nvidia's Vera Rubin GPUs.
- A broad shortage of HBM has given memory suppliers pricing power, which has materially improved Micron's recent financial results.
- Company guidance points to continued rapid revenue and profit growth, and the upcoming fiscal fourth‑quarter report will be the market's near‑term test of that outlook.
- Rising AI infrastructure costs are already prompting changes by customers and cloud providers, including Microsoft raising some AI prices, Anthropic changing token billing, and Uber exceeding its 2026 AI budget quickly.
- Because HBM feeds GPUs with the data needed for large models, sustained high memory prices could raise data‑center operating costs and slow or reshape AI deployments by hyperscalers and enterprise users.