VAST Data and AMD Expand Alliance to Power the Next Era of AI Inference
VAST Data has expanded its collaboration with AMD to help enterprises and AI cloud providers build high-performance infrastructure for the rapidly growing inference and agentic AI market.
The partnership combines the VAST AI Operating System with 6th Gen AMD EPYC processors, AMD Instinct GPUs, and AMD Pensando Pollara 400 AI NICs, creating an integrated platform designed to support large-scale AI training, inference, retrieval-augmented generation (RAG), and agentic AI deployments.
“The industry is discovering that inference is fundamentally a data problem. Success depends on how effectively organizations can bring data, compute, memory and intelligence together as a single system.”
John Mao, Vice President, Global Technology Alliances, VAST Data
As organizations move beyond model training toward operational AI systems, infrastructure demands are changing significantly. Industry leaders are increasingly focused on managing data, memory, context, and compute resources efficiently, particularly as AI agents and reasoning models require persistent context and real-time access to large datasets.
At the center of the collaboration is VAST’s Disaggregated Shared Everything (DASE) architecture, which enables unified access to data services, storage, databases, event streaming, and AI workloads across distributed environments. The architecture is designed to improve infrastructure utilization while supporting multi-tenant AI environments at scale.
A key aspect of the alliance includes the adoption of 6th Gen AMD EPYC processors across VAST’s next-generation infrastructure platforms. The new processors introduce PCIe Gen-6 support, delivering higher I/O bandwidth and lower latency for AI data services.
“Our expanded collaboration with VAST combines AMD EPYC CPUs and Instinct GPUs with the software foundation customers need to accelerate inference, improve infrastructure efficiency and deploy AI at scale.”
Derek Dicker, Corporate Vice President, Enterprise Business Group, AMD
The companies have also introduced a joint AI Infrastructure Reference Architecture alongside DriveNets, featuring AMD Helios rack-scale AI systems, VAST AI OS, and DriveNets networking fabric. The reference designs provide deployment guidance for training, inference, reinforcement learning, and KV-cache-intensive AI workloads.
According to VAST, early testing with AMD Instinct MI355X GPUs demonstrated up to 9x faster time-to-first-token and nearly 10x greater token throughput when using VAST’s KV-cache optimization capabilities.
The expanded ecosystem includes collaborations with partners such as TensorMesh, EmbeddedLLM, Core42, Crusoe, Vultr, TensorWave, Phanos.AI, and DriveNets, all focused on accelerating production-grade AI infrastructure.
As enterprises increasingly deploy AI-powered agents and inference-driven applications, VAST and AMD believe open, scalable, and data-centric architectures will play a critical role in delivering the performance, efficiency, and operational simplicity needed for the next generation of AI factories.
