Raghu Raghuram, VMware | VMware Explore 2023
summary
Claims VMware is betting on multicloud by letting teams bring the model to the data instead of moving petabytes around, arguing that GPU-heavy, multi‑GPU clusters plus high‑bandwidth storage and networking are the real cost drivers. Argues domain‑specific smaller models trained longer can match larger models cheaply; cites 100x wave in productivity and extraction of contract terms from PDFs as a concrete use case. Predicts 50% private infra for generative AI, warns of model provenance and data‑supply‑chain risks, and says Broadcom backing brings more R&D funding.
AI-written summary · about
chapters (10)
- 0:00Welcome and introduction
- 0:43VMware's multicloud strategy
- 2:47AI's impact on systems cost
- 4:32Gen AI's 100x productivity leap
- 6:59AI-native enterprise apps
- 8:47Public vs private AI infrastructure
- 10:57Private AI and legal compliance
- 12:00Organic AI innovation surge
- 13:41VMware's cross-environment advantage
- 14:43Closing and next chapter
