Sharon AI has signed a five-year strategic agreement with Rafay Systems to orchestrate AI infrastructure across its operations.
The arrangement is designed to support the orchestration of up to 150,000 GPUs over the term of the agreement, giving Sharon AI a common platform to manage infrastructure across multiple locations, tenants and workloads.
Sharon AI, an Australian neocloud operator listed on Nasdaq, will use Rafay's software as a central orchestration and operations layer across its AI Factory environments. The goal is to move from managing individual clusters to a more standardised operating model as the company expands in Asia-Pacific and other markets.
That model covers how computing infrastructure is provisioned, governed, monitored and delivered to customers. It also spans both Kubernetes and virtual machine environments through a single platform.
Platform shift
The agreement marks a shift in how Sharon AI plans to run its estate as it adds more GPU-based infrastructure. Instead of treating deployments as separate clusters, it is establishing a control layer above the underlying physical systems to apply common rules and processes across sites.
Rafay said its platform includes infrastructure lifecycle management, workload orchestration, governance and monitoring. That would allow Sharon AI to apply the same operating standards across different customer environments while keeping tenants isolated from one another.
The companies presented the deal as part of a broader effort to make AI infrastructure easier for customers to use without owning or operating the hardware themselves. That is increasingly central to the neocloud business model, in which providers rent access to scarce GPU resources and the surrounding software stack needed to run AI workloads.
Sharon AI said the standardised platform should also help it build internal operating practices and expertise as its footprint grows. It expects the system to support more automated service delivery, improve infrastructure utilisation and maintain common governance across deployments.
James Manning, Co-founder and Chief Executive Officer of Sharon AI, outlined the company's rationale for the deal.
"Sharon AI is delivering the trusted infrastructure and operational capabilities required to help customers realize measurable value," said James Manning, Co-founder and Chief Executive Officer of Sharon AI.
"Our agreement with Rafay gives us a standardized platform for orchestrating and managing AI infrastructure across our expanding footprint. This should enable us to manage our compute capacity more efficiently, accelerate service delivery and provide customers with a consistent and secure experience as we scale."
Growing demand
Demand for GPU infrastructure has risen sharply as companies seek computing capacity for model training, inference and other AI workloads. That has created opportunities for specialist cloud providers, but it has also increased the operational burden of managing hardware fleets across sites and customers.
In that environment, software layers that control provisioning, access, monitoring and governance have become more important. Providers must allocate resources between customers, maintain service reliability and enforce security and policy controls while trying to keep expensive assets in use.
Rafay's pitch to operators in that market is that a unified control plane can reduce the complexity of managing bare-metal systems, virtual machines and container-based workloads. For Sharon AI, the attraction appears to be consistency across a growing estate rather than reliance on separate tools or processes in each deployment.
Haseeb Budhani, Co-founder and Chief Executive Officer of Rafay Systems, said the deal reflects a broader need in the AI infrastructure market.
"Operating AI infrastructure at scale requires more than access to GPU capacity," said Haseeb Budhani, Co-founder and Chief Executive Officer of Rafay Systems.
"Providers also need the orchestration, automation and governance capabilities required to transform that infrastructure into secure, reliable and production-ready AI services. Sharon AI is building a highly scalable AI infrastructure platform for customers across Asia-Pacific and beyond, and we are proud to provide an operational foundation that can support its continued growth."
The agreement's reference to up to 150,000 GPUs over five years points to Sharon AI's intention to build for much larger deployments over time. While the figure does not indicate installed capacity today, it gives a sense of the operational scope the company wants its software layer to handle as demand grows.
For operators in the AI infrastructure market, that level of scale raises questions not only about access to chips and data centre space, but also about how quickly systems can be deployed, assigned to customers and supervised once in production. The Sharon AI-Rafay agreement is aimed as much at that operational challenge as at raw compute supply.