Anyscale on Azure, powered by Ray
Build and deploy AI with Ray in your own Azure tenant on AKS.
Use Azure-native security and pay with your existing consumption commitment.
Own your AI, on an open-source stack
Anyscale is powered by Ray, the open-source engine for distributed AI: data processing, training, and inference in Python. Anyscale runs it on AKS, the Kubernetes your team already operates. Your code stays on open-source Ray APIs.
+80% GPU utilization
Orchestrate multiple workloads on a shared resource pool with priority-aware workload scheduling across reserved or on-demand GPUs.
4x faster development
Developers use the open-source libraries they know, like PyTorch and vLLM. Anyscale handles cluster operations.
Runs in your tenant
Your data, models, and compute never leave your Azure subscription. Existing governance and access controls apply unchanged.
Customers
Empowering customers building the next generation of AI
Wayve uses Ray, and increasingly Anyscale on Azure to run distributed ML and data pipelines across large CPU and GPU fleets with improved efficiency and resiliency.”
Run and scale the complete AI lifecycle
Run any model, or any Python library with Anyscale on Azure
Unstructured data processing
Enhance model performance with scalable pipelines to curate text, video, images, documents, etc
Training & fine-tuning
Scale training from one to thousands of GPUs using your ML framework and technique of choice
ML & LLM model inference
Run an internal inference platform in your Azure tenant on Azure Kubernetes Service (AKS)
Embedding and search
Generate embeddings at scale, with unified pipelines with preprocessing on CPUs and inference on GPUs
LLM post-training & RL
Close the learning loop from data to training to inference on a single engine
Simulations and parallel runs
Submit isolated, production-grade runs through Anyscale Jobs to run multiple experiments in parallel
Industry Scenarios
Learn how AI leaders use Anyscale
Autonomy & Robotics. Train and deploy multimodal AI across LIDAR, sensor, iamge, and video data to power autonomous vehicles, physical robots, and real-time control systems.
Commerce & Retail. Run recommendation, ranking, and demand pipelines across text, image, and behavioral data to drive conversion and personalization at scale.
Fintech & Banking. Build and scale fraud detection, credit scoring, and prediction models on sequence and embedding data with speed and reliability.
Media, Gaming & Digital. Power batch inference and training pipelines for search, generative AI features, and content moderation.
Biotech & Life Sciences. Train and fine-tune protein folding and molecular design models (AlphaFold, Boltz, RFDiffusion) across sequence, structure, and imaging data.
Azure Native Integration
Built on the Azure services you already use
Identity, policy, and billing integrated with Azure services you already trust.
Azure Kubernetes Service
Ray workloads run on AKS in your subscription. Anyscale manages the cluster lifecycle.
Microsoft Entra ID
Single sign-on over OAuth with your existing identity provider. No separate identity store to manage.
Azure RBAC
Authorization with built-in roles, scoped to organizations, projects, and workloads.
Azure Policy
Your existing compliance policies apply to Anyscale resources, unchanged.
Azure Key Vault
Secrets and credentials stay in your Key Vault and are referenced by Anyscale workloads.
Azure Marketplace
Purchase on your Microsoft invoice. Spend counts toward your Azure consumption commitment.