SC25: MinIO Unveils ExaPOD Reference Architecture for Exascale AI

MinIO Inc., a player in data foundation for Enterprise AI, announced MinIO ExaPOD, a modular reference architecture for building and operating exascale AI.

Developed in collaboration with Intel, Solidigm, and Supermicro, ExaPOD is powered by MinIO AIStor – the Exascale Data Store for the AI Enterprise – and extends MinIO’s proven DataPOD design from 100-petabyte units to one-exabyte building blocks, giving enterprises and neocloud providers a repeatable blueprint for sustainable cost-efficient AI performance at scale.

As AI agents, long-context models, and multimodal workloads push data volumes from petabytes to multi-exabytes, traditional storage architectures have become the bottleneck. GPUs sit idle not for lack of compute, but because data doesn’t arrive to the GPU fast enough. ExaPOD eliminates these bottlenecks by reducing and stabilizing latency at exascale, ensuring a consistent, high-throughput data path that keeps AI workloads continuously fed and operational. Additionally, ExaPOD delivers predictable TCO and sovereign deployment on industry-standard hardware, avoiding egress surprises and cloud lock-in while meeting residency requirements.

“AI is not simply about adopting the latest model or GPU, but about re-architecting how data is stored, moved, and made accessible at scale,” said AB Periasamy, co-founder and co-CEO, MinIO. “The winners in the AI era will be defined by their ability to efficiently deliver data at exascale performance with hyperscaler economics. ExaPOD makes that possible, providing a simple, modular architecture for enterprises to build their own AI infrastructure on their terms, with complete control and no compromises.”

The validated ExaPOD stack integrates Supermicro high-density platforms, Intel Xeon 6 processors, and Solidigm enterprise SSDs with MinIO AIStor software to deliver the low-latency, high-throughput performance needed for AI training, inference, and continuous operations. Designed to operate like hyperscaler infrastructure but on-prem, ExaPOD unifies hardware and software-defined storage into a balanced, repeatable system that scales seamlessly across racks and data centers. ExaPOD sets a new standard for density and power efficiency, delivering a massive all-inclusive 36PiB usable capacity/rack unit to minimize data center footprint while consuming an average of 900 W power/PiB usable capacity (including cooling) to free up maximum power resources for GPU compute.

“ExaPOD combines MinIO’s high-performance data software with Supermicro’s density-optimized platforms to deliver balanced AI data performance,” said Vik Malyala, president and managing director, EMEA SVP, technology and AI, Supermicro. “Our PCIe Gen5-validated systems with optional liquid cooling maximize NVMe capacity while maintaining energy efficiency and serviceability, enabling consistent throughput for the most demanding AI data paths.”

“Enterprises are discovering that unlocking AI’s potential starts with breaking down data silos and accelerating data access,” said Chris Tobias, GM, Americas tech leadership, Intel. “This collaboration with MinIO, featuring Intel Xeon 6 processors, simplifies how customers move, store, and serve their data at scale, allowing them to focus on building value and driving innovation.”

“Enterprises leading in AI require infrastructure that scales with performance while keeping economics predictable,” said Avi Shetty, senior director of AI ecosystems and partnerships, Solidigm. “Our work with MinIO in Solidigm’s AI Central Lab has shown that the ExaPOD Reference Architecture, powered by MinIO AIStor and Solidigm eSSDs, delivers resilient, high-performance, and cost-efficient data access at exascale. This turns infrastructure from a constraint into a competitive advantage.”

The ExaPOD Reference Architecture reflects the industry’s accelerating demand for energy-efficient, high-density, sovereign AI infrastructure. It represents a shift from traditional storage to AI data stores, where data fluidity, speed, and control define competitive advantage. ExaPOD offers a clear, supportable path into the exascale AI era, aligning with MinIO’s mission to deliver AI without limits.

MinIO will showcase ExaPOD at SuperComputing ’25 in St. Louis, November 16–21, at booth #6513.

Resources:
ExaPOD Reference Architecture Whitepaper
Blog: Introducing MinIO ExaPOD: The Reference Architecture for Exascale AI