Browse Azure News (138)

Brendan Burns explains why Microsoft was named a Leader in the 2026 Gartner Magic Quadrant for Container Management, and uses the announcement to outline how AKS, Azure Container Apps, Azure Arc, and Fleet Manager are positioned for AI, hybrid, and edge container operations at scale.

Rena Liu announces Azure AI Speech LLM 2607 and explains what changes for developers building speech-to-text and real-time voice experiences, including better multilingual and mixed-language recognition, improved punctuation and numeric transcription, lower latency, and a simpler phrase list parameter for domain-specific customization.

Steve Sweetman explains how to govern AI agent spend in Microsoft Foundry by combining request-path token limits, observability (traces, monitoring, evaluations), and finance-grade budgets, then tying those signals to ROI so teams can decide what to optimize, scale, or retire.

Maria Bledsoe outlines how Azure IaaS resiliency is shifting from one-time DR planning to continuous design-and-operations practices, highlighting new and updated Azure capabilities for assessing resiliency posture, reducing blast radius during failures, and validating recovery readiness.

Juan Montes profiles Kinaxis’ Maestro platform and how it uses Azure (including AKS, Databricks, Azure OpenAI, Cosmos DB, and Azure AI Content Safety) to run large-scale supply chain scenario modeling, forecasting, and agentic AI experiences. The piece also touches on Kinaxis’ internal use of GitHub Copilot for an agentic, PR-driven development workflow.

Jakub Oleksy’s August 2026 GitHub availability report breaks down five production incidents across GitHub Actions, authentication, and Copilot services, including what failed, how GitHub mitigated impact, and what reliability work is planned next (capacity, autoscaling, retries, and regional failover).

Nick Brady rounds up July and August 2026 updates for Microsoft Foundry, including GA for Hosted Agents, Voice Live integration, and Toolboxes, plus Model Router pool/region changes and new model releases. It also calls out SDK migrations (Python/JS/Java/.NET) and runtime requirement bumps you should test before upgrading.

Microsoft Security Research and Lior Leizerovich introduce the Cloud Web Applications Threat Matrix, a MITRE ATT&CK-aligned framework for understanding and prioritizing threats against cloud-hosted web apps and serverless platforms, with practical examples spanning identities, deployment pipelines, managed runtimes, and connected cloud resources.

Mark Russinovich, Eric Henry and Sai Vaidhyanathan lay out a practical way to decide when an Azure workload should use two availability zones versus three, focusing on component-by-component requirements like capacity after a zone loss, durability targets, and quorum-based behavior rather than defaulting to “three zones everywhere.”

sbaynes announces codename MDASH, Microsoft’s multi-model agentic code-scanning system now deployed in Azure Government, aimed at finding and validating exploitable vulnerabilities across software supply chains with fewer false positives and more actionable results for security teams.

Aseem Datar explains how Microsoft Discovery applies agentic AI to scientific and engineering R&D, focusing on adaptive reasoning loops that explore multiple hypotheses, preserve evidence, and decide when to change strategy or involve domain experts. The post also highlights benchmark results for Discovery Engine with CLIO across several scientific domains.

davidwright walks through a practical, staged path for turning an Azure-hosted SaaS app into a transactable Azure Marketplace offer, using Git-Ape “skills” to generate version-controlled decision records for offer selection, multi-tenancy, landing zones, fulfillment/metering, security evidence, deployment blueprints, and Partner Center onboarding.

Shayak Lahiri explains why Edge AI shifts security responsibility into customer-owned infrastructure, and outlines practical patterns for protecting model weights, credentials, and data. The post focuses on gating sensitive asset release using runtime attestation and artifact provenance, plus deterministic mediation to constrain what models can do.

Justin Bettencourt’s August 2026 Azure SDK roundup highlights new and updated client libraries across .NET, Java, JavaScript/TypeScript, Python, Go, and more, including Document Translation 2.0.0, Azure Storage Blobs 12.30.0-beta.1, and the first stable release of Azure AI Discovery.

Wes Steyn and Eduard van Valkenburg announce a preview integration that adds durable, cross-session memory to Microsoft Agent Framework agents using Azure Cosmos DB. They explain how the Cosmos-backed context provider participates in the agent lifecycle, and show a Python example that stores conversation turns, extracts memories in the background, and recalls them in later sessions.

sbaynes announces GPT-6 Astra rolling out via the Microsoft Foundry Limited Access Program, focusing on agentic, multi-step work (planning, tool use, and computer use across apps) and the enterprise controls needed to run it safely in production, including identity, networking, governance, and monitoring.

Jeremy Winter argues that getting enterprise AI into production is less about picking a single model and more about running an end-to-end system on Azure—spanning infrastructure, data, governance, security, and operations. The post highlights Microsoft AI Foundry, Fabric, Purview, and Azure databases, plus modernization workflows that include GitHub Copilot.

Steve Sweetman and Naomi Moneypenny announce general availability of OpenAI’s GPT-6 Astra in Microsoft Foundry, focusing on agentic, cross-application “computer use” and enterprise readiness. The post highlights deployment options (Standard vs Provisioned Throughput), Global vs US Data Zone availability, token-based pricing, and the security/governance controls used to run these workflows in production.

Brent Halazon explains how Microsoft’s Physical Security Engineering Team used Azure Arc and Azure Virtual Desktop to standardize and scale hybrid operations across a global datacenter footprint, improving governance, observability, patching, and the operator experience without moving workloads fully into Azure.

Microsoft Security Research, Sagar Patil, Arlette Umuhire Sangwa, Jesse Birch and Ravikant Tiwari break down a hands-on-keyboard intrusion that starts with Microsoft Teams helpdesk impersonation, escalates through remote-control sessions and a malicious MSI, and ends with Node.js-based persistence and WinRM lateral movement, with concrete mitigations and Defender XDR hunting queries.

ryanjadams explains how to keep Fabric Data Factory pipelines and Copy Jobs fully private when moving data to or from Snowflake, using Workspace-level Private Link, a private Azure Storage staging account, and Snowflake’s Azure Private Link storage integration.

Jeff Hollan explains how “context engineering” in Microsoft Foundry can reduce enterprise AI agent costs without sacrificing quality, by tightening what goes into the context window each turn. The post breaks down practical levers—knowledge retrieval, tool selection, reusable procedures, and memory—plus governance controls like Entra ID and Purview.

Nimrod Shalit introduces a new set of Microsoft Learn resources for implementing CI/CD in Microsoft Fabric, covering platform concepts, best practices, and hands-on tutorials for deploying Fabric items across dev/test/prod using Git integration, Azure DevOps pipelines, Terraform, the Fabric CLI, and the fabric-cicd library.

bytecascade explains how Microsoft Fabric Eventstream can securely ingest private operational data (like factory MQTT telemetry) for real-time analytics by using an Azure virtual network bridge and a Fabric streaming virtual network data gateway, without exposing public endpoints.

Lily Ma explains how to extend Azure SRE Agent with operational tools by hosting Model Context Protocol (MCP) servers on Azure Connector Namespace. The post walks through deploying an Azure SQL MCP server with azd, wiring it into SRE Agent as an MCP connector, and securing access using managed identity.

sbaynes announces a Microsoft/AWS collaboration that streamlines private, high-performance connectivity between Azure and AWS using standardized OpenAPI-based interoperability, with an emphasis on operational simplicity, predictable bandwidth, and built-in security features like MACsec.

robece-msft continues the Microsoft Fabric Business Events series by outlining five real-world business event scenarios and the shared event-driven pattern used to turn business moments into trusted signals and downstream actions across Fabric.

sbaynes breaks down three concrete architecture choices PONS made to build a legal AI platform on Azure for regulated customers: separating public legal knowledge from private client data, leaning on managed Azure services to reduce ops overhead, and turning compliance requirements into enforceable technical controls.

Kristen Womack and Jeffrey C. summarize the August 2026 Azure Developer CLI (azd) releases, including the GA extension framework, new container deployment options for Azure Functions, breaking command/telemetry terminology changes, and a long list of provisioning, pipeline, and interactive UX fixes.

Lauro Ojeda compares self-hosted PostgreSQL with managed PostgreSQL services on Azure, focusing on what you gain and give up around control, operational effort, resilience, security, and cost predictability. It includes a practical checklist to help teams decide which operating model fits their requirements and constraints.

robece-msft continues the Microsoft Fabric Business Events series by showing how to react to business events using Fabric Activator and Eventhouse, focusing on practical event-driven patterns for turning event streams into signals and actions across Fabric.

Wes Steyn shows how to take a Microsoft Agent Framework “claw” from a working local prototype to something you can actually run for other people: one shared agent factory feeding multiple hosts, with OpenTelemetry observability, Purview governance, Foundry hosting, and repeatable evals for quality and regression checks.

Microsoft Security Research, Yash Gund and Sumith Maniath break down real-world attacks against exposed AI workloads (LiteLLM, RAGFlow, and Kestra), showing how gateways and orchestration layers become credential-rich control planes. The post maps intrusion chains, IOCs, and MITRE techniques, then provides concrete mitigations and Defender hunting queries.

murray-kp rounds up the August 2026 Microsoft Fabric updates across OneLake, Data Engineering, Data Warehouse, Real-Time Intelligence, Data Factory, and Fabric data agents, including Git-based CI/CD improvements, Spark runtime upgrades, new security controls, and expanded AI-driven conversational analytics capabilities.

Steve Sweetman breaks down four practical levers in Microsoft Foundry for reducing the cost of agentic AI in production: routing each request to the right model, caching repeated prompt content, optimizing prompts and agent configurations, and using observability plus evaluation to measure cost per outcome.

Dan Hellem announces the public preview of GitHub Copilot Code Review for Azure Repos and explains what’s changed since the technical preview, including new onboarding controls, support for Managed DevOps Pools, custom review instructions, automatic reviews via branch policies, and improved Azure Cost Management visibility.

Anna Hoffman introduces the Azure SQL Database Foundations series: four videos (with companion repos) that walk developers through getting started with Hyperscale, migrating and optimizing existing databases, scaling for performance, and adding AI features like vector search and RAG directly on operational data.

Nir Mashkowski announces a 30-day trial for Azure SRE Agent that removes baseline always-on charges during setup, plus GA VNet integration and a public preview of Live Reports for building repeatable operational views that refresh with live data from connected systems.

Bruno Capuano shows how to take a minimal Microsoft Agent Framework console app and deploy it as a Foundry Hosted Agent in Azure using one prerelease NuGet package, three lines of C#, and two azd commands—while getting managed identity, session state, observability, and an OpenAI-compatible /responses endpoint.

Vlad Fedorov shares GitHub’s write-up of the August 17 outage, what failed under peak traffic, and what the team is changing to reduce the chance and blast radius of future incidents. It covers capacity shortfalls, recovery actions, and concrete reliability work like retry limits, safer rollouts, and expanded Azure footprint.