Amphix Launches to Build AI Factories from the Ground Up

SEATTLE, WA – March 09, 2026 – A new consortium of infrastructure and software firms today launched The Amphix™ AI Infrastructure Platform, an ambitious initiative aimed at solving one of the most significant barriers to AI adoption: the staggering complexity of building and scaling AI production environments. The collaboration between RAVEL, Strata Expanse, and Available Infrastructure promises an integrated, end-to-end solution that guides enterprises from raw land selection to fully operational AI, debuting with a plan for 30 U.S. sites.

The platform's formal introduction is slated for the upcoming NVIDIA GTC 2026 conference, signaling a direct appeal to the ecosystem at the heart of the AI revolution.

Tackling the AI Infrastructure Gauntlet

Enterprises racing to harness the power of AI are frequently stopped in their tracks by a daunting set of logistical, technical, and financial hurdles. The journey from a successful proof-of-concept to a production-scale "AI factory" has proven to be a fragmented and perilous one. Industry analysis reveals that the primary obstacles are no longer just about algorithms or data, but the physical and digital foundation required to run them.

A core challenge is what some experts call the "Density Dilemma." Modern AI workloads, driven by power-hungry GPUs, can demand between 50kW and 100kW per rack—a tenfold increase over traditional data center racks. This has created an industry-wide scramble for power-ready sites and advanced liquid cooling solutions, as legacy facilities are often unable to cope with the thermal and electrical load. This pressure is compounded by the high cost of GPU hardware, complex software integration, and the persistent shortage of skilled talent capable of architecting these sophisticated systems.

"Deploying AI at scale has historically required stitching together land acquisition, power procurement, cooling systems, networking, hardware validation, and software orchestration across multiple vendors and environments," the companies noted in their announcement. Amphix aims to eliminate these constraints by unifying the entire stack under a single, repeatable model.

A Unified Platform from Land to Logic

The Amphix platform is built on a foundation of strategic partnerships, each providing a critical layer of the AI infrastructure stack. The goal is to create a seamless, pre-integrated environment that removes the guesswork and risk from building AI capabilities.

The physical layer is provided by Strata Expanse, which develops infrastructure-ready land complete with resilient power and cooling. The company's unique 'Gray Space as a Service™' subscription model allows customers to access this foundational infrastructure as an operational expense (OpEx) rather than a massive upfront capital expenditure (CapEx). This model, structured as a Real Estate Investment Trust (REIT), is designed to accelerate deployment by bypassing the lengthy and complex processes of site selection and construction.

"By integrating land, power, networking, and validated AI architectures, we’re enabling enterprises to build AI factories with confidence from day one,” said Ellen Taylor, Chief Revenue Officer at Strata Expanse.

Connecting these physical sites is a cybersecure, zero-trust network fabric from Available Infrastructure. This addresses the critical need for high-bandwidth, low-latency, and secure connectivity, which is essential for distributed AI workloads and protecting sensitive data. “A securely connected network of national sites is essential for scalable AI operations,” added Daniel Gregory, CEO of Available Infrastructure. “We’re proud to provide this key piece of the puzzle for Amphix.”

The entire system is managed by RAVEL's intelligent orchestration software, which acts as the platform's central nervous system. The software provides policy-driven, power-aware management of compute, storage, and networking resources, optimizing performance and efficiency across the distributed environment.

The Power of the Ecosystem and Validated Blueprints

Beyond the core partners, the strength of Amphix lies in its growing ecosystem of certified technology providers. This strategy ensures that customers are not just getting disparate components, but pre-validated, performance-tuned architectural blueprints. RAVEL and Strata Expanse announced that DDN, a leader in AI and HPC storage, and Supermicro, a top provider of high-performance servers, have joined as certified technology partners. Their systems, powered by technology from NVIDIA, AMD, and Intel, form the core of the platform's compute and storage offerings.

This ecosystem approach directly addresses a major pain point for IT leaders: the risk and time involved in validating hardware and software configurations. By offering certified solutions, Amphix promises a level of certainty that is often absent in custom-built AI infrastructure.

“The future of AI infrastructure isn’t just about GPUs,” stated Philippa Carroll, Chief Product Officer at RAVEL. “It’s about validated, scalable systems that connect physical capacity to production workloads seamlessly. Amphix delivers certainty across the entire journey.”

Further solidifying this promise, a new validated solution based on the DDN Enterprise AI HyperPOD™ architecture will be available for proof-of-concept deployments starting April 1, 2026. This architecture, built on Supermicro hardware and accelerated by NVIDIA's AI platforms, is designed for demanding workloads in sectors like financial services, life sciences, and retrieval-augmented generation (RAG).

Prove Before You Scale: The Rise of AI Centers of Excellence

Perhaps the most compelling component of the Amphix strategy is its network of AI Centers of Excellence (COEs). Following a successful pilot in Ohio, the platform is launching with a plan to establish these facilities at its 30 U.S. sites. These COEs function as real-world testing grounds, allowing organizations to validate their specific AI workloads under production-level power, thermal, and performance conditions before committing to a large-scale deployment.

This "prove before you scale" model is designed to dramatically reduce both technical and financial risk. Enterprises can use the COEs to benchmark performance, optimize their models, and ensure that their AI initiatives will deliver the expected value when moved into production. This approach directly confronts the reality that many AI projects which perform well in a lab environment fail to scale effectively, a key reason why many organizations remain stuck in the pilot phase.

By using the same orchestration and governance framework in the COEs that is used in full production, Amphix provides a consistent and predictable path from initial testing to a nationwide AI factory. This allows organizations to build with confidence, knowing their architecture has already been proven to work at scale. The strategic implementation partner, SourceCode, will support these deployment and expansion initiatives, ensuring customers have the expertise needed to transition smoothly from validation to full-scale operation.