VAST Data and AMD Deepen AI Infrastructure Alliance for Inference Workloads — VAST Data

Event summary

VAST Data expanded its collaboration with AMD to optimize AI infrastructure for inference workloads, leveraging 6th Gen AMD EPYC CPUs and Instinct GPUs.
The partnership introduces an AI Infrastructure Reference Architecture developed with DriveNets, supporting model training, inference, reinforcement learning, and KV cache workloads.
Early testing showed a 9X speedup in time-to-first-token (TTFT) and 9.7X more token throughput using VAST for KV Cache offloading with high concurrency agentic AI workloads.
The collaboration includes ecosystem partnerships with TensorMesh and EmbeddedLLM to accelerate deployment of production-ready inference architectures.

The big picture

As organizations transition from model training to operationalizing AI agents and inference services, the demand for efficient, scalable infrastructure is growing. This partnership between VAST Data and AMD addresses critical bottlenecks in data management, memory, and compute resources, positioning them as key players in the evolving AI landscape. The collaboration underscores a broader industry shift towards unified, high-performance systems that support complex, real-world AI applications.

What we're watching

Performance Scaling: How VAST Data and AMD's optimized inference architectures will scale across diverse enterprise environments.
Ecosystem Integration: Whether the expanded partnerships with TensorMesh and EmbeddedLLM can accelerate adoption of production-ready AI solutions.
Market Differentiation: The pace at which VAST Data can differentiate itself in the competitive AI infrastructure market through this collaboration.