Uploaded November 2026 | Updated November 2026, 2 days ago
Scaling enterprise AI workloads while managing computing costs, energy consumption, and infrastructure performance requires organizations to make informed technology investments. Explore how NVIDIA and Oracle are advancing AI infrastructure with Vera Rubin, open Nemotron models, and accelerated vector search to support agentic AI, inference, and enterprise data applications.
Don’t miss a thing from Oracle AI World 2026. Register for AI World On Air here: https://social.ora.cl/6003BGN9vJ
How can enterprises scale AI infrastructure while improving performance, controlling costs, and maximizing the value of their computing investments? In this partner interview on Oracle TV at Oracle AI World 2026, Dave Salvatore, director of Accelerated Computing Products at NVIDIA, discusses the expanding collaboration between NVIDIA and Oracle, including AI infrastructure announcements, open models, and accelerated database capabilities.
The conversation explores Oracle's plans to adopt NVIDIA's Vera Rubin and Vera CPU platforms and what these technologies could mean for organizations deploying AI at scale. Salvatore explains how the growing computing demands of agentic AI are changing data center design, with particular emphasis on maximizing AI performance per megawatt and improving the economics of AI inference.
A central focus is NVIDIA's approach to building AI infrastructure that is productive, durable, and versatile. Salvatore discusses the importance of delivering more AI tokens per dollar while maintaining infrastructure that can support workloads over time. He cites benchmark testing showing a 30-fold improvement in tokens per dollar compared with a previous-generation NVIDIA platform, illustrating the potential performance and cost benefits of newer AI computing architectures.
The interview also examines how organizations should evaluate AI infrastructure investments. Beyond performance and cost, Salvatore emphasizes versatility, explaining how NVIDIA accelerated computing on Oracle can support AI workloads alongside data processing, analytics, visualization, and scientific simulation.
Open AI models are another major theme. Salvatore introduces NVIDIA Nemotron, a family of open models available through the OCI model catalog. He discusses why open models can appeal to enterprises seeking greater flexibility and control over the intellectual property they develop using AI.
The conversation concludes with an example of full-stack collaboration between NVIDIA and Oracle. Salvatore discusses work to accelerate vector search in Oracle AI Database 26ai using NVIDIA's cuVS library. By supporting retrieval-augmented generation (RAG) applications, these capabilities can help organizations make enterprise data stored in Oracle databases more accessible to AI-powered applications.
Part of Oracle TV's coverage of Oracle AI World 2026, this interview connects NVIDIA accelerated computing, Oracle Cloud Infrastructure, AI inference economics, open models, and database acceleration with the practical challenges of scaling enterprise AI.
Like what you learned about Oracle AI World 2026? Subscribe now https://social.ora.cl/60025k87j
Follow us here:
LinkedIn: https://social.ora.cl/6000B6oR6p
X: https://social.ora.cl/6002B6oRDG
Instagram: https://social.ora.cl/60095eXDN
Facebook: https://social.ora.cl/60065enQp
Scaling enterprise AI workloads while managing computing costs, energy consumption, and infrastructure performance requires organizations to make informed technology investments. Explore how NVIDIA and Oracle are advancing AI infrastructure with Vera Rubin, open Nemotron models, and accelerated vector search to support agentic AI, inference, and enterprise data applications.
Don’t miss a thing from Oracle AI World 2026. Register for AI World On Air here: https://social.ora.cl/6003BGN9vJ
How can enterprises scale AI infrastructure while improving performance, controlling costs, and maximizing the value of their computing investments? In this partner interview on Oracle TV at Oracle AI World 2026, Dave Salvatore, director of Accelerated Computing Products at NVIDIA, discusses the expanding collaboration between NVIDIA and Oracle, including AI infrastructure announcements, open models, and accelerated database capabilities.
The conversation explores Oracle's plans to adopt NVIDIA's Vera Rubin and Vera CPU platforms and what these technologies could mean for organizations deploying AI at scale. Salvatore explains how the growing computing demands of agentic AI are changing data center design, with particular emphasis on maximizing AI performance per megawatt and improving the economics of AI inference.
A central focus is NVIDIA's approach to building AI infrastructure that is productive, durable, and versatile. Salvatore discusses the importance of delivering more AI tokens per dollar while maintaining infrastructure that can support workloads over time. He cites benchmark testing showing a 30-fold improvement in tokens per dollar compared with a previous-generation NVIDIA platform, illustrating the potential performance and cost benefits of newer AI computing architectures.
The interview also examines how organizations should evaluate AI infrastructure investments. Beyond performance and cost, Salvatore emphasizes versatility, explaining how NVIDIA accelerated computing on Oracle can support AI workloads alongside data processing, analytics, visualization, and scientific simulation.
Open AI models are another major theme. Salvatore introduces NVIDIA Nemotron, a family of open models available through the OCI model catalog. He discusses why open models can appeal to enterprises seeking greater flexibility and control over the intellectual property they develop using AI.
The conversation concludes with an example of full-stack collaboration between NVIDIA and Oracle. Salvatore discusses work to accelerate vector search in Oracle AI Database 26ai using NVIDIA's cuVS library. By supporting retrieval-augmented generation (RAG) applications, these capabilities can help organizations make enterprise data stored in Oracle databases more accessible to AI-powered applications.
Part of Oracle TV's coverage of Oracle AI World 2026, this interview connects NVIDIA accelerated computing, Oracle Cloud Infrastructure, AI inference economics, open models, and database acceleration with the practical challenges of scaling enterprise AI.
Like what you learned about Oracle AI World 2026? Subscribe now https://social.ora.cl/60025k87j
Follow us here:
LinkedIn: https://social.ora.cl/6000B6oR6p
X: https://social.ora.cl/6002B6oRDG
Instagram: https://social.ora.cl/60095eXDN
Facebook: https://social.ora.cl/60065enQp










