Anyscale, which is the company responsible for the open source Ray project and an AI compute platform innovator, has recently announced that it is offering the Anyscale on Azure preview service. This fully native integration, powered by Azure Kubernetes Service and Azure Resource Manager, enables companies to run AI workloads such as multimodal data processing and model training on their very own Azure infrastructure. The integration offers identical security, identity, billing, and operational workflows as standard Azure services, while helping organizations achieve up to 90% in cost savings.
For companies moving from experimental AI to production-ready applications, there is an increasing trend of seeing higher costs and regulations associated with using external third-party APIs of machine learning models. To combat this, businesses are opting to move towards building their own open-source models. This strategic pivot helps companies secure proprietary data assets and replace volatile per-token pricing with controlled, owned compute resources, ultimately transforming raw data into a compounding competitive advantage.
“AI has quickly become one of the largest and least predictable line items in the enterprise IT budget,” said Keerti Melkote, CEO of Anyscale. “The companies pulling ahead are not necessarily spending less on AI. They are gaining more control over how that spend scales. Instead of only renting intelligence through APIs, they are building and operating AI systems inside their own cloud. Anyscale is for the teams that have decided their AI is core enough to own. We give them one platform to curate massive multimodal datasets, develop their own models, and deploy those models at scale, all now possible on the Microsoft Azure infrastructure they already trust.”
“There’s growing interest from enterprise customers in building AI inside their own Microsoft Azure cloud environment, on their own data, with more control over how costs scale,” said Brendan Burns, Technical Fellow and CVP, Azure Cloud Native, Microsoft. “Anyscale on Azure brings the popular open-source Ray engine directly into Azure, giving customers a great option to build and operate AI systems within their existing Azure environments.”
Also Read: NVIDIA Launches First-of-Its-Kind ‘Vera’ CPU, Pioneering a New Architectural Era for Autonomous AI Agents
Geospatial AI provider Xoople exemplifies this industry shift by leveraging massive, proprietary satellite data to build localized intelligence. Anyscale on Azure provides the necessary infrastructure to accelerate their development cycles while keeping engineering teams focused on business outcomes rather than backend operations.
“With Anyscale on Azure, Xoople can reliably run massive AI workloads over planetary-scale satellite imagery, transforming complex spectral data into decision-ready intelligence,” said Milos Colic, VP of Engineering at Xoople. “Anyscale lets our teams focus on models and outcomes rather than infrastructure, dramatically accelerating the path from experimentation to deployment. For our product teams and theirs, this means a faster stream of information, more agility, and improved risk management.”
Autonomous vehicle innovator Wayve also utilizes the platform to aggregate vast GPU capacity across multiple regions, bypassing the infrastructure limits of single-region clusters.
“Wayve and Microsoft have a deep collaboration focused on scaling embodied AI and the infrastructure behind it. As Wayve’s AI platform and data operations have grown, Microsoft Azure has become a core part of its large-scale compute and ML stack. Wayve uses Ray, and increasingly Anyscale on Azure to run distributed ML and data pipelines across large CPU and GPU fleets, supporting large-scale inference, analytics, and dataset processing with improved efficiency and resiliency. This enables Wayve to train and deploy its autonomous driving AI at the speed and scale needed for safe, real-world deployment globally.” – Girish Venkataramani, VP of Engineering, Wayve AI
Streamlining the AI Lifecycle to Maximize Financial Efficiency
AI operational costs are rarely restricted to model utilization alone; they are heavily driven by fragmented architectures used across data curation, fine-tuning, evaluation, and inference. Historically, platforms combined CPU-heavy data pipelines with GPU-intensive hosted APIs, resulting in brittle frameworks, high overhead, and limited cost visibility.
This gap is bridged by Anyscale’s unique unified runtime engine that leverages Ray an open-source technology used to build out architectures at leading companies specializing in AI such as Cursor, Physical Intelligence, and xAI. Through consolidation of CPU and GPU tasks in a single Azure cloud ecosystem, businesses are able to improve their GPU efficiency, remove any potential data latency, and do away with uncertain API pricing. According to users, they experience a decrease in experimental times of up to 4x and 90% cheaper TCO.
Native Integration for Enterprise-Grade Sovereignty and Security
Developed via a deep technical partnership between Anyscale and Microsoft, the architecture ensures that data pipelines, training weights, and proprietary assets never breach the customer’s secure Azure tenancy.
Since provision of the service is done through Azure Resource Manager, each asset is monitored and controlled just like a first-party resource. Because of this, security teams are able to effortlessly enforce Microsoft Entra ID policies, apply role-based access controls (RBAC), and follow existing audit protocols. Thanks to this design, even the most regulated industries such as healthcare, finance, and government benefit by having their meticulously documented data residency and compliance rules equally apply to new AI pipelines without any extra steps.
Besides, as the platform can be reached directly through the Azure Portal, the costs related to consumption may be charged against the existing Microsoft Azure Consumption Commitments (MACC). Because of this, there is no need to go through separate procurement or legal approvals.
