Jeff Denworth, VAST Data | Databricks Data + AI Summit 2023
summary
Claims that VAST is the first enterprise distributed storage to support NVIDIA DGX SuperPOD and that customers plan to process up to 10 exabytes drive the talk. It describes coupling massively parallel GPU farms (thousands of GPUs) with storage that serves hundreds of petabytes, aims to democratize hyperscaler-class SuperPODs, and previews an August unveiling of a "thinking machine" platform. The built integrations are already running with hyperscalers and target global distributed model training.
AI-written summary · about
chapters (10)
- 0:00Introduction and welcome
- 1:47Rethinking storage as data platform
- 3:51Data’s critical role in AI
- 5:35Infrastructure for deep learning
- 7:22Real-time data processing
- 8:45Future of data platforms
- 10:51Performance and cost in AI
- 12:13Upcoming platform reveal
- 13:57Advice for entrepreneurs
- 15:20Future of AI models
related talks
14mJohn Mao, VAST Data | Nvidia GTC 2024
12mAnthony Dina, Dell Technologies | Nvidia GTC 2024
17mScott Baker, IBM | Nvidia GTC 2024
20mPriyank Patel, Cloudera | Nvidia GTC 2024
28mScaling agentic AI: From infrastructure to enterprise outcomes
19mScalable Data and AI Solutions with Cloud-Native Architecture
12mKV Cache and Long-Context Inference: Solutions for Inference at Scale
10mKeynote Analysis