SharonAI Holdings Inc. (NASDAQ: SHAZ) announced on September 4, 2026, that it has entered into a five-year strategic agreement with Rafay Systems. Under the terms of the deal, Sharon AI will utilize the Rafay platform as a centralized orchestration and operations layer across its AI Factory environments.
The Rafay platform is designed to standardize how Sharon AI provisions, governs, monitors, and makes accelerated computing infrastructure available to customers. The agreement formalizes Sharon AI’s use of the platform as it moves from individual cluster deployments to a consistent operating framework intended to support growth across the Asia-Pacific region and globally.
The Rafay platform is intended to provide the automation, observability, governance, and secure multi-tenancy capabilities required to operate advanced AI infrastructure at scale. It is designed to help Sharon AI manage compute capacity efficiently and respond to different customer workload requirements.
James Manning, Co-founder and Chief Executive Officer of Sharon AI, stated that the agreement provides a standardized platform for orchestrating and managing AI infrastructure across the company’s expanding footprint. He noted that this should enable the company to manage compute capacity more efficiently and accelerate service delivery.
Haseeb Budhani, Co-founder and Chief Executive Officer of Rafay Systems, commented that operating AI infrastructure at scale requires more than access to GPU capacity. He stated that providers also need orchestration, automation, and governance capabilities to transform infrastructure into secure, reliable, and production-ready AI services.
The agreement is also expected to support the development of Sharon AI’s internal capabilities and expertise around the Rafay platform. The collaboration is intended to strengthen Sharon AI’s ability to automate service delivery, improve infrastructure utilization and reliability, and maintain consistent governance.
The architecture established through the agreement is designed to support the orchestration of up to 150,000 GPUs over the five-year term.
