VAST Data and AMD Expand Partnership to Accelerate AI Inference

What Happened

VAST Data and AMD announced an expanded collaboration on July 24, 2026, combining VAST Data's AI Operating System with AMD's EPYC processors, Instinct GPUs, and Pensando Pollara 400 AI networking cards to deliver what the companies describe as breakthrough inference performance. The partnership targets enterprises scaling AI deployments beyond model training into large-scale inference operations.

What Was Announced

The joint solution integrates the VAST AI Operating System with three distinct AMD hardware lines: AMD EPYC CPUs for general compute, AMD Instinct GPUs for accelerated AI workloads, and AMD Pensando Pollara 400 AI network interface cards for high-throughput data movement. The companies stated the combined stack is designed to address the performance and efficiency demands of the inference era, a phase in AI deployment characterised by serving trained models to end users and applications at scale rather than building those models.

No specific benchmark figures or pricing details were included in the announcement released via GlobeNewswire.

Background

VAST Data is a data storage and management company that has positioned its platform around AI infrastructure workloads. The company's AI Operating System is designed to manage data pipelines and storage at the scale required by large AI deployments. AMD is the second-largest producer of x86 server processors and competes directly with Nvidia in the AI accelerator market through its Instinct GPU line. The Pensando Pollara 400 is AMD's AI-focused network interface card, intended to reduce data movement bottlenecks between storage, memory, and compute.

The announcement comes as enterprise demand for inference infrastructure grows alongside the broader deployment of large language models and other AI systems across commercial applications. Industry analysts and vendors have increasingly drawn a distinction between the capital-intensive phase of training AI models and the ongoing operational costs of running those models in production, a phase that places different demands on hardware and software architecture.

What It Means in Practice

The collaboration means customers can deploy a validated, pre-integrated stack from two vendors rather than assembling and testing components independently. The VAST AI Operating System is positioned to manage data flow across AMD's compute and networking hardware, with the stated goal of reducing latency and improving throughput for inference queries.

The Pensando Pollara 400 AI NIC is specifically cited as a component enabling high-speed data transfer between storage and GPU resources, a bottleneck that becomes more pronounced as inference workloads require rapid access to large datasets or model weights.

The agreement represents an expansion of a prior relationship between the two companies. Neither party disclosed the financial terms of the collaboration or specified which customer segments or geographies are prioritised in the initial rollout.

Industry Context

The inference infrastructure market has attracted increasing attention from hardware vendors, software platform providers, and cloud operators as AI adoption moves from pilot projects to production systems. AMD has been actively expanding its ecosystem partnerships to broaden adoption of its Instinct GPU line, which competes with Nvidia's H-series and B-series accelerators in data centre deployments. VAST Data has similarly sought to position its platform as purpose-built for AI data management rather than general-purpose enterprise storage.

The expanded collaboration was disclosed through a joint press release distributed via GlobeNewswire on July 24, 2026.

What Happens Next

The companies have not announced a specific availability date for jointly validated configurations, but VAST Data and AMD are expected to present further technical details through partner and customer channels in the months following the announcement.