Arun Subramaniyan, Intel | Supercloud 4
summary
Generative AI delivers much faster and larger ROI than prior AI work, with a recent MIT/Wharton/BCG survey of ~600 companies reporting 11% in production (two+ cross-department projects in critical apps) versus ~8–10% historical ROI. Two clear use cases emerge: knowledge discovery (retrieval + generation, vector DBs and embeddings) and augmented services/co‑pilots that ground outputs with domain models. Most enterprise data stays on‑prem — roughly 95% of data remains dark — driving specialized on‑prem models, edge inference, and X86 CPUs with embedded AI acceleration.
AI-written summary · about
