Pure Storage® (NYSE: PSTG), the IT pioneer offering the world’s most advanced data storage technology and services, has just announced new validated reference architectures for running generative AI use cases, including a new reference architecture validated for NVIDIA OVX. As a leader in AI, Pure Storage, in collaboration with NVIDIA, provides customers worldwide with a proven framework to manage the high-performance data and computing requirements needed for successful AI implementations.
Highlights
Building on this collaboration with NVIDIA, Pure Storage offers the latest technologies for today’s businesses to address the rapid growth in AI demand. The new validated designs and proof of concepts include:
Retrieval-Augmented Generation (RAG) Pipeline for AI inference: To enhance the accuracy, timeliness, and relevance of Inference functionalities for large Language Models (LLM), Pure Storage has created an RAG pipeline using NeMo Retriever microservices and NVIDIA GPUs, and Pure Storage for fully flash enterprise storage. As a result, Pure Storage accelerates the information retrieval timelines for enterprises using their own internal data for AI training, all while ensuring the use of their latest data and eliminating the need for constant LLM training.
Certified NVIDIA OVX Server Storage Reference Architecture: Pure Storage has achieved OVX Server Storage validation, providing customers and channel partners with flexible storage reference architectures validated against key parameters to provide a solid infrastructure foundation for cost-effective and high-performance AI hardware and software solutions. This validation offers additional options to AI customers and complements Pure Storage’s certification for NVIDIA DGX BasePOD announced last year.
Vertical RAG development: To accelerate successful AI adoption in vertical sectors, Pure Storage is creating specific vertical RAGs in collaboration with NVIDIA. Initially, Pure Storage has created an RAG solution for financial services to summarize and query massive datasets more accurately than standard LLMs available on the market. Financial institutions can now gain insights faster using AI to create instant summaries and analysis from various documents and sources. Additional RAGs for healthcare and the public sector will also be launched.
Increased investment in the AI partner ecosystem: Pure Storage is investing more in its AI partner ecosystem with NVIDIA and has established new collaborations with Independent Software Vendors like Run.AI and Weights & Biases. Run.AI optimizes GPU usage through advanced orchestration and scheduling, while Weights & Biases’ AI development platform enables AI teams to develop, evaluate, and govern the model development lifecycle. Additionally, Pure Storage works closely with AI-focused distributors and service partners, including ePlus, Insight, WWT, and others, to make joint AI deployments more operationally efficient for customers.
Executive Insights
“Pure Storage has recognized the growing demand for AI from the outset and has provided an efficient, reliable, and high-performance platform for the most advanced AI implementations. Thanks to our long-standing collaboration with NVIDIA, the latest AI reference architectures and validated generative AI proof of concepts have become essential components for global enterprises looking to unravel the complexities of the artificial intelligence puzzle.” – Rob Lee, Chief Technology Officer, Pure Storage.
“NVIDIA’s AI platform reference architectures enhance with Pure’s simple, efficient, and reliable data infrastructure and offer comprehensive solutions to companies grappling with the complexities of AI, data analytics, and advanced computing. Pure’s latest validated reference architectures and proof of concepts are helping foster innovation in AI and provide companies with the tools needed to unlock new opportunities and drive transformative results.” – Bob Pette, Vice President of Enterprise Platforms, NVIDIA.
“At WWT, technological innovation is the foundation that allows us to deliver value services to customers worldwide. The demand for AI continues to grow, so we are thrilled to continue collaborating with two of the best providers in our field of testing. Pure Storage’s high-performance AI storage infrastructure and NVIDIA’s technological expertise in AI computing generate impressive synergies, enabling our joint customers to seamlessly incorporate artificial intelligence into their business operations and achieve significant return on investment.” – Ryan Avery, High-Performance Storage Architect, World Wide Technology.
“Our mission at Run.AI is to be the engine of the AI revolution, leading positive change in various sectors and enhancing the human experience. We focus our efforts on providing the essential software layer that makes AI infrastructure possible, so we are excited to collaborate with Pure Storage and NVIDIA to help drive more widespread (and successful) adoption of advanced enterprise AI initiatives.” – Yael Dor, VP, Global Sales and Partnerships, Run.AI.
“At Weights & Biases, we are dedicated to developing the best tools for machine learning and generative AI — including automating experimentation and tracking model artifacts, optimizing model evaluation and prompts, and enabling comprehensive governance and reproducibility of machine learning workloads. Our expanded collaboration with Pure Storage and NVIDIA empowers AI developers to develop and deploy superior models quickly, reliably, and efficiently.” – Seann Gardiner, VP of Business Development, Weights & Biases.
“Pure Storage’s validated NVIDIA reference architectures and proof of concepts provide companies across many sectors with a shortcut to AI success. Instead of spending valuable time and resources building an AI architecture from scratch, Pure’s proven frameworks help reduce the risk of costly project delays and guarantee a high return on investment in AI equipment expenses, such as GPUs.” – Mike Leone, Senior ESG Analyst.
Importance for the Industry
Currently, most AI implementations are scattered across different fragmented data environments — from the cloud to inappropriate storage solutions (often inherited). However, these fragmented environments cannot support the performance and network requirements needed to drive AI data pipelines and unleash the full potential of enterprise data.
Companies are ramping up AI adoption to drive innovation, optimize operations, and gain a competitive edge, making the demand for robust, high-performance, and efficient AI infrastructure greater than ever.
A pioneer in enterprise AI deployments, especially among a fast-growing set of Fortune 500 enterprise customers, Pure Storage provides a simple, reliable, and efficient storage platform for companies to fully leverage the potential of AI while reducing the associated risks, costs, and energy consumption.