Azure Weekly

Issue 571

12th July 2026

Email newsletters are tricky. There are so many different email clients, on so many different platforms, it's very difficult to please everyone. I did want to point out that if you struggle with the email newsletter (for example Gmail has a habit of truncating the email half way through), then we also have a web version of the newsletter, which offers a superior reading experience (fully responsive) and also offers the ability to toggle link summaries on and off - as I know some of the readers really like skimming titles, whereas others love the summaries to figure out if the longer article is worth their time. Head over to the website to give it a try.

Highlights for this week include:

  • Microsoft Foundry Model Deployment Pricing Update by Chris Hoder - Microsoft Foundry will raise prices for EU and non-US deployments by 10-50% starting September 1, 2026, and introduce an APAC Data Zone.
  • Built to bounce back: How Azure resiliency evolved by Rochak Mittal, Amit Ganguli, Surya Sripathi Raju, Molina Sharma - Azure's resiliency model focuses on a shared responsibility approach combining infrastructure foundations such as Availability Zones with customer-defined application architectures to ensure adaptable, recoverable, and trustworthy systems under real-world conditions.
  • External key management for Azure Managed HSM is now in public preview by Salim Chawro - External key management for Azure Managed HSM allows organizations to maintain encryption keys on owned hardware outside Azure datacenters, fulfilling stringent regulatory demands while retaining robust sovereignty via a FIPS 140-3 Level 3 HSM.
  • GigaWiper: Anatomy of a destructive backdoor assembled from multiple malware by Microsoft Threat Intelligence - GigaWiper merges multiple malware families into a Go-based backdoor capable of physical disk wiping, ransomware-style encryption without key storage, and multi-pass secure wiping, reflecting a trend toward modular destructive tools for enhanced efficiency and impact.
  • A Paradigm Shift in Cloud Operations with Azure SRE Agent by Nir Mashkowski - Azure SRE Agent reduces incident triage time for companies like Zafin, Provation Medical, and InEight from hours to minutes by automating evidence collection, issue classification, and action recommendations in cloud operations.
  • Generally Available: Support 5x churn in Azure Site Recovery by The Azure Updates Team - Azure Site Recovery now supports up to five times higher change rates, delivering 500 MB/s per VM and ensuring reliable disaster recovery for high IOPS workloads.

The new Fabric Weekly Newsletter launched on the 30th June, and we're going to stop covering the official Fabric site, to try and reduce the size of the newsletter If you're interested in all things Fabric, sign-up now!. and also don't forget about Power BI Weekly! (and don't forget subscriptions are "double-opt-in" - keep an eye out for an email which requires you confirm your subscription before you'll receive an issue). One Fabric-related post worth covering is:

  • Microsoft Fabric Workspace Topology Patterns by James Broome - Understanding workspace topology patterns in Microsoft Fabric is crucial for setting up secure, scalable, and manageable environments from the start, avoiding costly refactors later.

🤖 AI

  • Kernels Behind the Wall: Custom Python 3.10-3.14 Jupyter Kernels for Air-Gapped Azure ML This guide demonstrates building custom Python 3.10-3.14 Jupyter kernels for air-gapped Azure ML instances using conda-pack, private endpoints, and offline installation to overcome repository and kernel limitations.
  • AI ethics: Why opting out isn’t always the ethical move Opting out of AI ethics discussions misses opportunities to address critical issues such as water usage, job displacement, and copyright problems through informed engagement.
  • Monitoring & Observability in Microsoft FoundryPart 2: Configuration and Operations Monitoring & Observability in Microsoft Foundry Part 2 covers configuring Application Insights, operating in Azure Monitor, and setting up alerts for maintaining AI agent accuracy, safety, efficiency, and consistency.
  • Understanding Token Cost Anatomy: The Five-Headed Bill Nobody Budgeted For The article explains that AI spending can be broken down into five distinct cost categories-API calls, model compute, data ingestion, prompt engineering, and storage-helping teams understand where their budget goes beyond just total costs.
  • Available today: OpenAI’s GPT-5.6 in Microsoft 365 Copilot OpenAI's GPT-5.6 integrates into Microsoft 365 Copilot across Word, Excel, PowerPoint, Chat, and Cowork, boosting multi-step knowledge work with enhanced reasoning while upholding enterprise security and compliance.
  • Microsoft Foundry Model Deployment Pricing Update Microsoft Foundry will raise prices for EU and non-US deployments by 10-50% starting September 1, 2026, and introduce an APAC Data Zone.
  • Tokenomics | The new AI currency & your options explained Learn how to optimize AI token usage with strategies like compressing conversation history, capping output tokens, caching static context, and using tools such as Agent Optimizer and Model Router in Microsoft Foundry for smarter app design and reduced spend.
  • GCast 220: Creating a Workflow in Microsoft Foundry GCast 220 explains how to build sophisticated, multi-agent applications in Microsoft Foundry using its Workflow tool, enabling complex automation with no coding required.
  • Foundry IQ is now in Copilot Studio: Bring your enterprise data to every agent conversation Foundry IQ's integration into Copilot Studio boosts AI agent performance by 54%, offering enterprise-ready features such as Customer-Managed Keys, ACLs, and compliance certifications for secure and governed data access.
  • Who actually blocks AI in healthcare (and why) Healthcare AI deployments frequently stall due to structural barriers such as physician autonomy, works council surveillance concerns, patient privacy worries, legal liability fears, and CISO security vetoes, which are often not addressed by standard readiness checklists.
  • Before You Ship Your Agent: A Five-Step Path to Evaluations You Can Trust Before shipping your AI agent, ensure you have multi-turn evaluated sessions, a clear rubric defining "good," simulated real-user conversations, sampled production traces for comprehensive coverage, and benchmarked standards to continuously validate quality.
  • Multi-Agents in RavenDB RavenDB's multi-agent system enables creation of specialized agents coordinated via simple SubAgents configurations, reducing complexity and maintaining strict control.
  • Introducing Kimi K2.7 Code in Microsoft Foundry Kimi K2.7 Code on Microsoft Foundry boosts long-horizon coding efficiency by cutting token usage by 30%, while providing enterprise-level security, quicker deployment, and smooth integration with GitHub Copilot and VS Code.
  • Matthew Renze: AI Changes - Episode 409 Matthew Renze, a nine-time Microsoft MVP in AI and founder of Renze Consulting, discusses the evolution of AI integration-from simple assistant interactions to autonomous agency leadership-highlighting tools like OpenClaw, Hermes, and PaperClip AI.
  • Token Economics: The New FinOps for Agentic AI Token economics manages costs in agentic AI by using caching, tiered pricing, and context compression to lower expenses without sacrificing performance.

🔎 Analytics

  • Generally Available: Confidential Computing support for Azure Event Hubs Dedicated Azure Event Hubs Dedicated introduces Confidential Computing, using hardware-based TEEs to protect sensitive streaming data from unauthorized access during processing.
  • Generally Available: Open AI GPT-5.6 on Azure Databricks Open AI’s GPT-5.6 is now generally available on Azure Databricks, enabling secure model serving through the Model Serving Endpoint for seamless integration with Microsoft Foundry workflows.
  • Creating Microsoft Fabric Sandbox Capacities Chatbot Microsoft Fabric introduces time-based ephemeral sandbox capacities, allowing organizations to safely explore its features without worrying about costs or long-term resource commitments.
  • Microsoft Fabric Workspace Topology Patterns Understanding workspace topology patterns in Microsoft Fabric is crucial for setting up secure, scalable, and manageable environments from the start, avoiding costly refactors later.
  • The Data Engineering Patterns Series The Data Engineering Patterns Series explains how software engineering principles apply to data engineering, while also highlighting the complexities and challenges of that transition.
  • Fabric Data Agents can choose Query Language based on Context Fabric Data Agents dynamically choose the optimal query language-SQL, Kusto, DAX, or Ontology-for each context to enhance performance and integrate historical analytics with real-time data in one AI chat interface.
  • Optimising DAX: Model Design Comparisons and Testing Tips The article finds that the star schema offers best performance in DAX due to efficient compression of small dimension tables and medium-cardinality relationships, whereas flat tables can achieve good compression through run-length encoding despite having more columns.

🖥️ Compute

🚢 Containers

🗄️ Databases

🛠️ Developer tools

  • TypeScript 7 is done TypeScript 7 adds a native Go-compiled compiler for 8x-12x faster builds but lacks an official API, restricting integration with popular frameworks until version 7.1.
  • Claude AI Gets Yet Another Boost in VS Code 1.128 Visual Studio Code 1.128 enhances Claude AI with advanced agent workflows, attachment support, browser integration, shortcut enhancements, and improved enterprise telemetry management.
  • How GitHub Copilot enables zero DNS configuration for GitHub Pages GitHub Copilot CLI automates DNS setup for GitHub Pages, enabling developers to publish custom-domain websites with HTTPS in under 15 minutes without manual DNS record editing.
  • Things that Have Worked for Our OSS Community The author shares lessons learned from 15 years of OSS community work, emphasizing responsiveness, user-centric documentation, and automation to foster a thriving Critter Stack ecosystem.
  • GitHub's former CEO launches a distributed Git network built for the agentic coding age The company behind GitHub is creating a distributed Git network designed to support the emerging era of coding agents, enhancing infrastructure for autonomous development workflows.
  • Microsoft Testing Platform and .Net 10 The new Microsoft Testing Platform (MTP) introduced with .Net 10 provides a lightweight, portable alternative to VSTest, removing dependencies and improving test execution in CI pipelines, Visual Studio, and VS Code.
  • Avalonia UI vs Microsoft’s .NET MAUI Joseph Tomkinson's comparison shows MAUI has stronger mobile support and more third-party components while Avalonia UI leads in cross-platform consistency, desktop performance, and Linux support, with recent funding doubling project builds over six months.
  • Microsoft Agent Framework for .NET (part 1): what it is, how it works, and when to use it The Microsoft Agent Framework for .NET provides a scalable solution for developing AI agents with better integration than alternatives like Semantic Kernel, LangChain, and AutoGen.
  • Jeff Fritz on Copilot for App Modernization Jeff Fritz explains how GitHub Copilot’s Modernization Agents use AI to analyze legacy .NET and Java apps, plan upgrade strategies, and automate the execution process.

🔩 DevOps

  • How to automate Fabric data agent testing with promptfoo and GitHub Actions This guide demonstrates how to automate testing of Fabric data agents using promptfoo and GitHub Actions, ensuring every code change is validated with automated LLM tests.
  • How GitHub gave every repository a durable owner GitHub now requires a designated owner for each new repository, archives inactive ones, and uses custom properties to track ownership linked to services, teams, or individuals, enhancing security workflows and cutting manual work.
  • Automating cross-repo documentation with GitHub Agentic Workflows GitHub Agentic Workflows automates cross-repo documentation by generating a markdown workflow that prompts an AI agent to draft and open pull requests in the docs repository, ensuring timely updates without granting unrestricted write access.
  • GitHub availability report: June 2026 GitHub's June 2026 report details infrastructure progress, handling 100% of anonymous pull request reads with new services, increasing repository traffic to 50%, offloading 500,000 queries per second from the primary database, and enhancing access controls and capacity.
  • Introducing ROI for agents in Foundry ROI for Agents in Foundry connects agent performance data with business value metrics, enabling organizations to quantify whether production agents deliver measurable cost savings or revenue growth.
  • Q1 2026 Innovation Graph update: Open source collaboration is accelerating worldwide The Q1 2026 Innovation Graph update reveals a 16% quarter-over-quarter increase in outbound collaboration across economies, driven by heightened developer activity on platforms like GitHub.
  • Optimizing GitHub Copilot Cost in the Usage-Based Billing Era Optimizing GitHub Copilot cost involves selecting appropriate models, limiting token usage through context windows and session length, implementing budget controls, and segmenting consumption by user role to maintain value without unpredictable expenses.
  • You Can’t Govern What You Can’t Name Naming data pipelines creates audit trails, enabling proof of execution, authorization, and intent-essential for governance and trust in data operations.

🧬 Hybrid + multicloud

🎭 Identity

🔌 Integration

🎓 Learning and Certifications

⚖️ Management and Governance

🚌 Migration

  • Generally Available: Support 5x churn in Azure Site Recovery Azure Site Recovery now supports up to five times higher change rates, delivering 500 MB/s per VM and ensuring reliable disaster recovery for high IOPS workloads.
  • Golden Paths Are a Product. Treat Them Like One. Treating golden paths like products-by applying product discipline such as internal NPS tracking and SLOs-ensures they remain adopted, trusted, and continuously improved across multiple teams.
  • Built to bounce back: How Azure resiliency evolved Azure's resiliency model focuses on a shared responsibility approach combining infrastructure foundations such as Availability Zones with customer-defined application architectures to ensure adaptable, recoverable, and trustworthy systems under real-world conditions.

🌐 Networking

🔐 Security

📦 Storage

💻 Virtual Desktop Infrastructure

Get Azure Weekly in your inbox every Sunday

Sign up to receive the email every week and keep on top of all the announcements.

Don't forget to confirm your subscription. We have a double opt-in sign up process to prevent spam.

We will never provide your details to any third parties. We don't spam.

Azure Weekly Logo