Argentum AI picks Rafay for GPU software orchestration

Argentum AI has selected Rafay Systems to support its global GPU infrastructure platform, with the agreement focused on software orchestration for customer-specific compute environments.

The arrangement is designed to let Argentum provision, configure and manage separate software environments for each customer through a single control plane.

Argentum secures and deploys AI infrastructure across data centre sites ranging from 10MW to more than 100MW, with access to more than 3GW of power capacity across the US, Europe and other markets. It supplies GPUs, managed services and orchestration for customers including hyperscalers, neoclouds and large enterprises that need dedicated compute resources.

At this scale, infrastructure providers face the challenge of maintaining distinct baseline software stacks for different customers. As portfolios expand across multiple sites and regions, that task becomes increasingly difficult if each environment is managed separately.

Rafay says its platform provides Argentum with a single software layer across these deployments while preserving separate environments for individual offtakers. The aim is to reduce operational complexity as Argentum expands its footprint.

Market demand

The deal comes as spending on AI infrastructure continues to rise. IDC forecasts global AI infrastructure spending will reach USD $758 billion by 2029, while its tracker found spending on AI compute and storage hardware rose 166 per cent year on year in the second quarter of 2025 to USD $82 billion.

Gartner has also projected worldwide end-user spending on AI-optimised infrastructure-as-a-service will total USD $37.5 billion in 2026. The figures point to strong demand for GPU-backed infrastructure for model training and inference.

For Argentum, the software layer is tied to a broader business model built around financed deployments. Each deployment is financed through a special purpose vehicle structure backed by Blackstone, J.P. Morgan and Goldman Sachs, with customers dealing through a single counterparty under one service-level agreement.

The arrangement also reflects how AI infrastructure providers are trying to shorten deployment times while meeting different customer requirements. Argentum offers a turnkey model intended to take projects from contract to live infrastructure within weeks.

Haseeb Budhani, Chief Executive Officer and Co-Founder of Rafay Systems, described Argentum as an expansion of the vendor's customer base within the AI infrastructure market.

"Rafay has built a strong track record helping organizations deploying AI compute at massive scale bring scalable cloud management to their GPU fleets. Argentum represents an important expansion of that capability portfolio, as they operate at a foundational layer in the stack, connecting data center capacity and power to the largest AI operators in the world. They require the same kind of multi-tenant, configurable software orchestration that we have already perfected for cloud builders, so this partnership demonstrates the breadth of use cases Rafay can serve across the AI infrastructure ecosystem, supporting infrastructure intermediaries who make GPU capacity accessible at global scale just as effectively as we support the token factories building at the top of it," Budhani said.

Growth plans

The partnership also comes as Argentum prepares to extend its model. It plans to build its own neocloud offering on the same software platform rather than change infrastructure systems later.

That positions the Rafay platform not only as a tool for current deployments, but also as part of Argentum's expansion from infrastructure broker to cloud service provider. For operators in this segment, avoiding a technology migration later can reduce cost and disruption.

Andrew Sobko, Chief Executive Officer of Argentum AI, said the company needed a software partner that could match the scale of its infrastructure ambitions.

"Argentum's mission is to be the fastest, most flexible provider of turnkey GPU and power capacity for the world's largest AI operators, and Rafay is helping us accelerate our growth. We have built a global footprint of high-capacity data center sites and the operational expertise to deliver customized compute environments at scale to each customer, precisely when they need them. What we required was a software orchestration partner capable of matching that ambition, and Rafay gives us the tools to deliver a distinct, tailored software environment to every offtaker in our portfolio without multiplying operational overhead. With Rafay, we can continue leading this market and deliver the scale and precision demanded by the world's top AI infrastructure buyers," Sobko said.