Anyscale Launches Native Integration on Microsoft Azure for Enterprise AI Workloads

In a press release, Anyscale announced the public preview of Anyscale on Azure, a native integration that allows enterprises to run large scale AI workloads entirely within their own Microsoft Azure environment. The platform is built on Azure Kubernetes Service and Azure Resource Manager, enabling organizations to manage AI systems with the same security, billing, and identity controls as other Azure services.

Anyscale on Azure allows companies to build and deploy their own AI models on infrastructure they control, keeping proprietary data and assets within their cloud tenancy. The company reports that customers can achieve up to 90 percent cost savings compared to using external model APIs. The integration supports multimodal data processing, training, and inference, all managed through the Azure portal.

The solution runs entirely inside a customer’s Azure tenancy, ensuring that data, training pipelines, and model weights remain within the customer’s account. It inherits Azure’s security and identity framework, including Microsoft Entra ID policies and role-based access controls. This design supports enterprise requirements for data sovereignty and governance, particularly in regulated industries.

Early adopters such as Xoople and Wayve are using Anyscale on Azure to run large scale geospatial and autonomous driving AI workloads. The platform’s unified compute foundation aims to reduce infrastructure complexity and improve GPU utilization while maintaining cost efficiency.