SchumannGE details the latest AKS Arc developments from Ignite 2025, focusing on hybrid Kubernetes, AI model deployment, enhanced edge and disconnected support, and new security features for the AI-enabled enterprise.

AKS Enabled by Azure Arc: Powering Hybrid AI Applications from Cloud to Edge

Author: SchumannGE

Microsoft Ignite 2025 showcased significant updates to Azure Kubernetes Service (AKS) enabled by Azure Arc, marking a leap forward in hybrid, edge, and AI workload capabilities. This summary distills the key announcements, technical improvements, and their implications for organizations building and scaling AI-driven solutions across diverse environments.

Key Announcements at Ignite 2025

  • AKS on Azure Local Disconnected Operations (Public Preview): Run AKS clusters fully offline in sovereign or regulated environments, with asynchronous synchronization for updates.
  • AKS Local Small Form Factor Bare-Metal (Private Preview): Deploy AKS to compact edge hardware for industrial, retail, and factory scenarios, supporting optional GPU acceleration.
  • AKS on Windows Server & Azure Local Medium Improvements: Enhanced reliability, expanded hardware support, improved image management, certificate handling, and error detection.
  • 2-Node High Availability: Makes robust, production-grade AKS deployments feasible in space- and budget-constrained edge sites.
  • AI Foundry Local Integration: Provides offline AI model catalog access, local fine-tuning (including LoRA/QLoRA), and optimized execution for GPUs, NPUs, and CPUs directly in customer environments.
  • KAITO for Model Serving: Automates packaging, optimization, and deployment of AI models (ONNX, Hugging Face, or custom) for multi-modal inferencing across cloud, datacenter, and edge.
  • Edge RAG (Retrieval-Augmented Generation): Enables compliant, low-latency generative AI grounded in on-premises data via Azure Arc extensions.
  • AKS Container Apps Public Preview: Edge-native microservices and event-driven AI endpoints graduating to production-readiness on Azure Local.
  • Arc-enabled Azure Monitor Pipeline: Enhanced telemetry for hybrid/disconnected sites with local caching and later synchronization.
  • Security Improvements: KMS v2 for secrets encryption at rest, Workload Identity Federation for secret-less Edge identity, integrated Key Vault extension for secure storage.
  • Expanded GPU and Storage Support: NVIDIA RTX 6000 Ada, L-series GPU previews, GPU partitioning, and persistent storage enhancements for AI/ML/data workloads.

Technical Deep Dive

Disconnected and Hybrid Operations

Enterprises can now deploy, manage, and upgrade AKS clusters without continuous Azure connectivity. This benefits regulated and air-gapped scenarios, supporting critical workloads in isolated manufacturing, defense, and telecom environments. Temporary sync enables updates without internet persistency.

AI Development at the Edge

AKS Arc establishes itself as Azure’s platform for distributed hybrid AI workloads:

  • AI Foundry Local enables local training and fine-tuning of foundation models, delivering a complete offline AI dev loop. Models can be packaged for AKS clusters, even in air-gapped environments.
  • KAITO offers streamlined deployment for ONNX, Hugging Face, and custom models, optimized for various hardware.
  • Edge RAG supports end-to-end generative AI using proprietary on-premises data, maintaining data sovereignty and reducing cloud latency.
  • Expanded GPU Support helps organizations run resource-intensive AI/ML inferencing and training where data lives, leveraging new GPU types and partitioning techniques.

Application and DevOps Enhancements

  • AKS Container Apps simplify microservices and event-driven deployments at the edge, using existing Azure development models.
  • Arc Gateway streamlines hybrid cluster connectivity and onboarding, reducing firewall configuration complexity.
  • Operational improvements: Enhanced monitoring, secret management, and governance tools for distributed fleets.

Security Improvements

  • KMS v2 delivers better at-rest encryption for Kubernetes secrets, aligning with Azure’s native practices.
  • Identity Federation & Key Vault Extensibility improve cluster workload security without explicit secrets. Workloads can use federated identities and integrated Key Vault storage, ideal for complex regulatory contexts.

Observability and Storage at Scale

  • Azure Monitor Pipeline supports disconnected operations, capturing and syncing telemetry efficiently.
  • Azure Container Storage supports AI/ML workloads’ storage needs, with performance improvements for vector stores and embedding caches.

Conclusion

Microsoft’s 2025 Ignite announcements cement AKS enabled by Azure Arc as the most versatile and secure platform for running modern—especially AI-powered—applications across cloud, on-premises, and edge. The platform now supports offline development, robust security, GPU-powered inferencing, and easy, consistent management, enabling organizations to build, deploy, and operate next-generation AI and cloud-native solutions wherever needed.

This post appeared first on “Microsoft Tech Community”. Read the entire article here