About AMPHIX
A network of containerized AI factories. Distributed renewable power. A repeatable path to scale. Discover how DDN helped transform an ambitious infrastructure project into a production-ready model.
Summary
AMPHIX, managed by RAVEL and powered by Strata Expanse, is a new deployment model for AI infrastructure designed to provide a seamless path from physical site selection to fully operational AI production. By incorporating DDN Enterprise AI HyperPOD, a packaged solution from DDN, built by Supermicro, accelerated by NVIDIA technologies, and powered by DDN Infinia, AMPHIX provides a validated path from experimentation to production.
- 98.8% GPU utilization sustained for 24 hours
- Zero Thermal throttling or GPU errors
- 37°C Thermal headroom at peak load
A New Approach to AI Infrastructure
As demand for AI infrastructure continues to grow, organizations face increasing challenges securing the resources needed to deploy AI at scale. Access to power, facilities, compute infrastructure, and operational control has become a significant barrier to adoption.
AMPHIX’s founding partners recognized an opportunity to address this challenge. Through Strata Expanse’s portfolio of sites with access to power resources, and RAVEL’s deep expertise in AI infrastructure and workloads, the partners envisioned an energy-first, modular approach designed to support AI deployments wherever appropriate energy and infrastructure resources are available, including renewable and alternative power sources. By bringing compute closer to power, AMPHIX saw the potential to make enterprise AI infrastructure more accessible, scalable, and easier to operationalize.
What began as a vision for RAVEL to showcase AI infrastructure capabilities on a single Strata Expanse site with the help of a long-term collaboration with Supermicro, quickly evolved into something larger. Together the teams at Strata Expanse and RAVEL developed a broader vision for AI infrastructure that would connect over 100 sites, and through its Centers of Excellence help organizations evaluate solutions, validate architectures and orchestration policies, and establish a clear path from proof of concept to large-scale production.
The Challenge for AMPHIX
Transforming a Vision into a Repeatable Blueprint
While AMPHIX’s vision was compelling, bringing it to life required more than infrastructure resources alone.
Organizations exploring AI often face two interconnected challenges: determining how to move from experimentation to production and making the right infrastructure decisions along the way.
Selecting compute, storage, networking, and software platforms independently can introduce risk, slow deployment timelines, and make it difficult to create a repeatable path to scale.
AMPHIX recognized an opportunity to help simplify that process. And set out to create a model that would enable organizations to evaluate AI infrastructure, validate solutions against real-world workloads, and establish a foundation they could scale with confidence.
QuoteWe want to help customers de-risk their path from concept design through production deployment. Strata supports AMPHIX in this mission by providing a secure network of reliable, power-ready sites for deploying AI at scale.
Greg WoodSenior Director, Strata Expanse
To achieve that vision, AMPHIX required a technology ecosystem capable of delivering every layer of the AI infrastructure stack through proven, integrated solutions rather than custom-built environments.
An Ecosystem Built for Scale
To transform the AMPHIX vision into a deployable reality at its first Center of Excellence in California, RAVEL assembled an ecosystem of best-in-class technology leaders to deliver the full spectrum of AI infrastructure needed.
At the center of the deployment is DDN Enterprise AI HyperPOD, a turnkey AI factory built using Supermicro servers and accelerated by NVIDIA technologies. Delivered as a pre-integrated, pre-validated solution, HyperPOD combines the infrastructure, software, and data platform organizations need to deploy production-ready AI without months of design, integration, and testing. At its core is DDN Infinia, the data engine for AI inference, agents, and AI-native applications. Combined with RAVEL’s Orchestrate AI software and the broader AMPHIX framework, the solution provides a repeatable blueprint for deploying AI at scale.
Rather than requiring organizations to source, integrate, test, and validate every technology independently, HyperPOD provides a pre-validated architecture that aligns with the AMPHIX mission of simplifying AI adoption and accelerating the path to production.
QuoteAMPHIX helps customers move from proof of concept to production without having to build and integrate every layer of infrastructure themselves. For Strata, that also means creating a path to deploy AI infrastructure responsibly, with energy, resource use and community impact considered from the beginning.
Greg WoodSenior Director

Within the Ecosystem
- Strata Expanse provides the land, power, cooling, connectivity and supporting infrastructure that form the physical foundation of AMPHIX.
- Supermicro delivers the AI-optimized compute infrastructure that powers the HyperPOD architecture.
- NVIDIA provides the accelerated computing, networking, and AI platform technologies that enable AI training, inference, and data processing across the HyperPOD architecture.
- DDN Infinia provides the AI-native data platform and data engine for AI inference, agents, and AI-native applications.
- RAVEL provides a streamlined operational intelligence layer designed to continuously optimize AI infrastructure and workloads through its Orchestrate AI software platform, as well as the operational expertise that bring ecosystem solutions into AMPHIX and helps organizations manage and operationalize AI infrastructure environments.
- AMPHIX and its Centers of Excellence provide the collaborative AI Factory framework that enables organizations to evaluate, validate, deploy, and scale AI infrastructure solutions.
- Together Strata Expanse and RAVEL deliver the long-term vision behind the AMPHIX model, and deliver what’s needed for organizations to scale AI operations commercially across its sites.
The first deployment of HyperPOD at the AMPHIX Center of Excellence in California demonstrates the potential of containerized AI infrastructure, currently connected to the CIE grid with a future goal of incorporating distributed energy resources including solar, wind, and geothermal power. The deployment marks an initial step toward AMPHIX’s broader vision of bringing compute closer to available energy and developing a model for future AI factory deployments.

The DDN Difference
HyperPOD Provides the Foundation for Scalable AI Infrastructure
DDN Enterprise AI HyperPOD, a packaged solution from DDN, built by Supermicro and accelerated by NVIDIA, serves as the foundation of the California AMPHIX Center of Excellence deployment. As AMPHIX set out to build a network of AI Centers of Excellence, it needed an established, proven solution that could be deployed quickly and provide a repeatable foundation for customer proof-of-concepts. Rather than requiring organizations to assemble and validate the AI stack themselves, HyperPOD delivers a modular, pre-integrated, pre-validated AI factory that can be ordered in pre-defined sizes and deployed rapidly. Built on the NVIDIA AI Data Platform architecture and powered by DDN Infinia, the solution provides a production-ready foundation for AI inference, agents, and AI-native applications, helping organizations accelerate time to value, reduce deployment complexity, and de-risk critical infrastructure decisions.
QuoteHyperPOD stood out to us because it represents what we believe AI infrastructure needs to become: a packaged, production-ready solution built through collaboration across the ecosystem. By bringing together DDN, NVIDIA, Supermicro and RAVEL, we can validate the complete infrastructure stack against real-world workloads, not just individual components. At AMPHIX, we’ve built and tested this integrated environment and are now ready to make it available for customer POCs, giving organizations the opportunity to test their workloads, validate their configuration and build confidence before moving into production.
Philippa CarrolChief Product Officer, RAVEL
As AI adoption grows, infrastructure performance depends on more than compute alone. Data must move efficiently between storage and GPUs to support training, inference, retrieval-augmented generation (RAG), and analytics workloads at scale. Without the right data architecture, bottlenecks can limit performance and reduce the value of infrastructure investments.
Within HyperPOD, NVIDIA provides the accelerated computing and AI platform technologies that power AI workloads, while DDN Infinia delivers the AI-native data platform that keeps data flowing efficiently to those resources. Together, they help maximize GPU utilization, reduce latency, and create a high-performance foundation for enterprise AI operations.
For AMPHIX, that foundation supports a broader goal: creating a repeatable deployment model that can scale across future locations. By combining Supermicro infrastructure, NVIDIA technologies, and DDN Infinia within a validated architecture, HyperPOD gives organizations a consistent starting point for deploying AI infrastructure with greater confidence and predictability.
That confidence is backed by real-world validation. RAVEL stress-tested HyperPOD across both inference and training workloads. In the 24-hour training test, HyperPOD sustained 98.8% GPU utilization, with zero thermal throttling or GPU errors and 37°C of thermal headroom at peak load. These results give partners and customers performance evidence to inform their evaluations before making infrastructure decisions.
QuoteWe wanted proven technology partners that could help customers get to production faster, and DDN was a natural choice because of its leadership in AI data infrastructure.
Greg WoodSenior Director, Strata Expanse
The result is a faster path from experimentation to production. Rather than spending months selecting, integrating, and validating infrastructure components, organizations can focus on deploying applications, operationalizing AI, and delivering business value.
By combining NVIDIA AI Data Platform with DDN’s AI-native data platform, HyperPOD helps AMPHIX deliver its vision of a scalable infrastructure model designed to support long-term AI growth.
Ready to test it yourself? Organizations can apply for a proof of concept at an AMPHIX Center of Excellence to validate HyperPOD against their own workloads and requirements before moving into production.
