At Pure Accelerate 2019, IT organizations learned how they can help their businesses bring AI development out of the shadows and into an “AI-first” mindset.
Most organizations and their IT leaders want to lean into embracing AI, positioning IT as an enabler rather than an inhibitor. This week, we announced important capabilities that will make it simpler for every enterprise to develop their best AI-powered applications faster — and deploy them in production at-scale sooner.
To get there, we’re making it easier for data scientists to develop models with greater iterative speed and, ultimately, maximum business impact. At the same time, we’re continuing to make it easier for organizations to access world-class AI-ready infrastructure facilities that ease and accelerate deployments. It’s a win-win for everyone involved in turning data into business insights at enterprise scale.
AI Data Hub
AI Data Hub is an end-to-end AI data pipeline — spanning initial exploration and prototyping to model training and inference — from Pure Storage that’s powered by NVIDIA GPUs, systems and software. By enabling accelerated movement of massive amounts of data through every phase of the development workflow, AI Data Hub can help organizations break down data storage silos associated with legacy architectures.
NVIDIA supercharges the AI Data Hub architecture, beginning with our RAPIDS suite of data science libraries built on CUDA-X AI, to deliver GPU-accelerated data ingest, manipulation and model training. AI Data Hub uses Pure Storage AIRI, built on NVIDIA DGX systems, to offer the fastest performance for training with multi-system scale. And it deploys effortlessly on NVIDIA T4 servers running inference.
AIRI-as-a-Service
In addition to streamlining AI development, Pure and NVIDIA are removing a fundamental implementation roadblock faced by many customers whose data centers aren’t AI-ready.
Extending the successful model introduced by the DGX-Ready Data Center Program, we’re partnering with Pure on the new AIRI-as-a-Service offering. This taps into a network of proven DGX colocation providers to offer a spectrum of services ranging from hosting customer-owned AIRI infrastructure to delivering AIRI-as-a-Service in a utility consumption model.
The offering will help customers of any size deploy AI infrastructure sooner by eliminating the burden of transforming their data centers to support the unique facilities demands of AI compute and affordably offering the capacity they need.
Learn more at the links below: