Azure in June 2026: Build Brings HorizonDB and Azure Linux, RI Deadline Is One Day Away, and Product Terms Get Rewritten

Executive Summary

June 2026 was shaped by Microsoft Build (June 2-5), a cluster of infrastructure retirements, and a significant rewrite of Microsoft Product Terms governing AI services. Three things need attention before June ends:

• The Azure Reserved VM Instance retirement deadline is July 1, 2026 – tomorrow as of this writing. Any RI on the affected series (Av2, Amv2, Bv1, D, Ds, Dv2, Dsv2, F, Fs, Fsv2, G, Gs, Ls, Lsv2 for 1-year; Dv3, Dsv3, Ev3, Esv3 for both terms) that expires on or after July 1 will not renew. Teams that have not yet traded into Azure Savings Plans for Compute or purchased new RIs on supported series must act today.
• Azure HorizonDB entered public preview at Microsoft Build. It is a fully managed, PostgreSQL-compatible database purpose-built for cloud-scale AI applications. For FinOps teams, this introduces a new database pricing tier alongside existing Azure Database for PostgreSQL Flexible Server. Teams evaluating database commitments should understand the pricing differences before purchasing new reservations.
• Microsoft Product Terms were substantially rewritten in June 2026 for the third consecutive month. A new umbrella definition pulls all Azure AI services under a single compliance framework. The Enterprise AI Services Code of Conduct now applies to every model deployed through Microsoft Foundry, including third-party models. Teams with AI governance programs need to review the updated terms before deploying new Foundry models.

For FinOps teams, three additional changes matter: Azure Linux entered public preview at no licensing cost, Azure Blueprints service is being retired, and granular cost controls for tenant-level, group-level, and user-level billing were announced for Azure Databricks Genie (moving to pay-as-you-go pricing July 6, 2026).

All pricing figures in this document require verification at azure.microsoft.com/pricing before acting on them.

Azure Reserved VM Instance Retirement: July 1, 2026 Is Tomorrow – Last Chance to Act

The July 1, 2026 cutoff for Azure Reserved VM Instance purchases and renewals on legacy series was covered in last month’s issue. As of June 30, one day remains. This section covers what to do in the final hours and what happens after the deadline passes.

Confirmed Affected Series and What Changes July 1

The following VM series lose RI purchase and renewal availability on July 1, 2026, per the Microsoft Learn transition guide (learn.microsoft.com/azure/cost-management-billing/reservations/manage-legacy-vm-reservations-after-july-1-2026 – confirmed live):

• 1-year RIs only: Av2, Amv2, Bv1, D, Ds, Dv2, Dsv2, F, Fs, Fsv2, G, Gs, Ls, and Lsv2 series.
• Both 1-year and 3-year RIs: Dv3, Dsv3, Ev3, and Esv3 series.

Existing reservations continue to provide discounts through the end of their current term. What stops: any new purchase, any renewal, and any auto-renew for these series. Auto-renew does not override the retirement – if an RI on an affected series expires on or after July 1, the auto-renew attempt will be blocked and the VM reverts immediately to pay-as-you-go rates.

What Microsoft Recommends: Three Options

Option	What It Does	Best For	Action Before July 1
Trade in for Azure Savings Plan for Compute	Exchanges current RI for prorated credit toward a Savings Plan covering any compute	Teams expecting VM family changes, scaling, or modernization	Submit trade-in in Azure portal under Reservations
Renew RI before July 1 (final window)	Purchases a new RI on the affected series for one final term before the cutoff	Teams with stable workloads not planning modernization near-term	Purchase before midnight UTC June 30
Migrate workloads to newer series + new RI	Moves VMs to Dv5, Ev5, or equivalent and purchases new RIs on supported families	Teams planning performance upgrades alongside commitment renewal	Begin migration now; purchase new RI post-migration

The trade-in path is Microsoft’s primary recommendation for flexibility. The final-renewal path is available until midnight UTC June 30 for teams committed to staying on the legacy series for one more term. Verify trade-in mechanics at learn.microsoft.com/azure/cost-management-billing/savings-plan/reservation-trade-in – terms change.

For a full comparison of Azure Savings Plans versus Reserved VM Instances including flexibility, savings rates, and when each mechanism delivers better return, the Usage.ai guide on commitment-based discounts across Azure covers the decision framework in detail.

Microsoft Build 2026 (June 2-5): Azure HorizonDB, Azure Linux, and Fabric Updates

Microsoft Build 2026 ran June 2-5 with an emphasis on agentic application development. The Azure-relevant launches with FinOps and infrastructure implications:

Azure HorizonDB: Public Preview – New PostgreSQL-Compatible Database Tier

Azure HorizonDB entered public preview at Microsoft Build. It is a fully managed, PostgreSQL-compatible database that combines PostgreSQL familiarity with cloud-scale architecture purpose-built for AI applications. Per the official Azure Blog post (azure.microsoft.com/blog/microsoft-build-2026-building-agentic-apps – confirmed live), HorizonDB is designed for high-scale, mission-critical AI applications where standard Azure Database for PostgreSQL Flexible Server architecture presents limitations at AI workload scale.

For teams evaluating database commitments, HorizonDB introduces a new pricing tier distinct from Azure Database for PostgreSQL Flexible Server. HorizonDB pricing and commitment eligibility are separate from Azure Database for PostgreSQL Flexible Server during preview. Verify current reservation and savings plan eligibility before making commitment decisions.

HorizonDB also includes Microsoft Defender for Cloud integration in preview, providing continuous security and compliance assessments for teams with compliance-driven database deployments.

Azure Linux: Now in Public Preview at No Licensing Cost

Azure Linux entered public preview at Microsoft Build. It is a secure, cloud-optimized Linux platform built by Microsoft for consistency and control at scale, offered at no licensing cost. It is designed to reduce operational overhead for teams managing patching, security, and compliance across Azure infrastructure.

The no-licensing-cost positioning is the key FinOps angle. Teams currently paying for RHEL or SUSE Linux licenses on Azure VMs should evaluate whether Azure Linux meets their workload and compliance requirements. If it does, migrating eligible workloads to Azure Linux can eliminate Microsoft-paid OS licensing costs for supported workloads, subject to application compatibility and support requirements.

Microsoft Fabric and Database Announcements at Build

Microsoft Build included several Microsoft Fabric and database announcements that affect Azure data workloads:

• Azure Cosmos DB and Fabric integration updates – tighter integration between Cosmos DB and Microsoft Fabric for agentic application data pipelines.
• PostgreSQL discovery and assessment tooling – new tools to evaluate Oracle and PostgreSQL environments and provide readiness insights, sizing guidance, and cost estimates for migration to Azure Database for PostgreSQL. Relevant for teams planning Oracle-to-Azure migrations.
• Microsoft Fabric SKU Estimator updated – the Fabric SKU Estimator was updated to help teams size Fabric capacity purchases. Teams running analytics workloads on Fabric should re-run the estimator if they have not done so recently before purchasing or renewing capacity reservations.

Microsoft Product Terms June 2026: AI Services Compliance Framework Rewritten

Microsoft updated its Product Terms for the third consecutive month in June 2026, with substantial changes to the AI services compliance framework. Source: SAMexpert analysis of the June 2026 Product Terms update (samexpert.com/microsoft-product-terms-june-2026/ – confirmed live). Teams with AI governance programs, legal, and procurement should review these changes before deploying new Foundry models.

New Umbrella Compliance Framework for All Azure AI Services

A new umbrella definition in the June Product Terms pulls all Azure AI services under a single compliance framework. Previously, individual model-provider terms (for Grok, Llama, Mistral, Black Forest Labs) were embedded directly in the Product Terms as separate sections. The June update consolidates these into an external documentation page. The Enterprise AI Services Code of Conduct now applies to every model deployed through Microsoft Foundry, including third-party models.

Practical impact: if your organization has compliance sign-offs on specific Foundry model deployments, those sign-offs may need to be re-reviewed against the new Code of Conduct umbrella. The structural change does not affect costs or entitlements directly, but it does change the compliance surface for any organization using third-party models through Foundry.

Azure Hybrid Benefit for Azure Local: Narrowed Scope

The June Product Terms insert one new sentence into the Azure Hybrid Benefit section: Azure Hybrid Benefit for Azure Local is available only for hyperconverged deployments with cloud connect management. Per the SAMexpert analysis:

• Only Windows Server Datacenter qualifies (Standard licences do not qualify for Azure Local, though they remain eligible for Azure VM AHB).
• The benefit is restricted to customers on an Enterprise Agreement with active Software Assurance, or a Cloud Solution Provider subscription.

Teams running Azure Local deployments should verify whether their license type qualifies under the narrowed terms before counting Azure Hybrid Benefit savings in their Azure Local cost model.

Foundry Models Renaming: Azure Direct Models Becomes Foundry Models Sold by Azure

The term Azure Direct Models has been replaced throughout the Product Terms with Foundry Models sold by Azure, reflecting the continued rebranding of the Microsoft AI platform. Azure AI Studio became Azure AI Foundry at Ignite 2024, became Microsoft Foundry in January 2026 Product Terms.

The June update extends the Foundry branding to the models themselves. The same renaming cascades through Agent 365 and Power Platform. This is a naming change with no cost impact, but any internal documentation, procurement contracts, or compliance policies that reference Azure Direct Models will need to be updated.

Microsoft Discovery Added to Limited Access Services Register

A new service called Microsoft Discovery was added to the Limited Access Services register without description or documentation in the June Product Terms. Limited Access Services require registration and are restricted to customers managed by a Microsoft account team. The SAMexpert analysis flags this as an undocumented addition.

Teams with enterprise Azure agreements should confirm with their Microsoft account team what Microsoft Discovery covers if they encounter it in billing or service registration.

Compute and Infrastructure: Azure Blueprints Retirement, Confidential Live Migration, and AVS AV36 VCF BYOL

Azure Blueprints Service Retirement

Microsoft has announced the retirement of Azure Blueprints and recommends migrating to Azure Deployment Stacks and Azure Policy. Teams using Azure Blueprints for environment governance and policy assignment should migrate to Azure deployment stacks and Azure Policy assignments, which are the recommended replacement capabilities. Azure Blueprints was a governance service for defining repeatable sets of Azure resources with policy assignments, role assignments, and ARM templates.

Cost implication: Azure Blueprints itself was billed through the underlying resource deployments it managed, not as a standalone service. Migration to deployment stacks does not change the cost of the resources being governed. The operational impact is on any automation or pipeline that calls the Blueprints API directly.

Confidential Live Migration for Intel TDX VMs

Azure can now perform live migration of Intel TDX (Trust Domain Extensions) virtual machines without restarting them, while preserving cryptographic protections during platform servicing. This feature is specifically relevant for workloads in finance and healthcare that require both continuous availability and confidential computing guarantees.

Previously, some platform maintenance operations for TDX-backed VMs could require a restart, depending on the maintenance event. Confidential Live Migration removes that requirement. Availability depends on specific hardware and region support. Confirm compatibility for your VM sizes and regions before relying on this capability for uptime SLA planning.

NVv3-Series VM Retirement: September 30, 2026

Microsoft announced the retirement of NVv3-series VM sizes on September 30, 2026. The affected sizes are Standard_NV12s_v3, Standard_NV12hs_v3, Standard_NV24s_v3, Standard_NV24ms_v3, Standard_NV32ms_v3, and Standard_NV48s_v3. These are GPU-accelerated VMs used for visualization workloads.

Teams running workloads on NVv3 have until September 30, 2026 to migrate. The recommended migration target is NVadsA10 v5-series or NVv5-series VMs, which provide better performance per dollar for visualization and VDI workloads. If NVv3 workloads are covered by Reserved Instances, check whether trade-in or migration to a supported series before expiry is available.

AVS AV36 VCF BYOL 1-Year RI Availability Ends June 30, 2026

Azure VMware Solution AV36 VCF BYOL 1-year Reserved Instances are available only until June 30, 2026. After that date, AV36 VCF BYOL 1-year RIs can no longer be purchased (confirmed from October 2025 Microsoft announcement). Teams running AV36 nodes under VCF BYOL licensing must plan their reservation strategy before this date. The AV36 node type itself retires on June 30, 2028, giving teams a transition window, but the RI purchase option closes much earlier. AV36P, AV48, AV52, and AV64 remain available.

Storage and Files: Azure Files Gets First-Class Resource Status, Azure Migrate Adds File Assessment

Azure Files: New microsoft.FileShares Namespace Makes Shares First-Class Resources

Azure Files introduced a new file share-centric management model under the Microsoft.FileShares namespace, reaching general availability in June 2026. File shares are now first-class Azure resources independent of the storage account structure, simplifying automation, policy application, and managed identity integration.

The practical change for operations teams: file share lifecycle management, access policies, and RBAC assignments can now be managed directly through Azure Resource Manager without navigating through the parent storage account. This simplifies automation in large deployments where hundreds of file shares previously required storage-account-level policy workarounds.

Cost implication: the new resource model does not change file storage pricing. However, teams with automation that billed storage-account-level operations to cost centers may need to update their tagging and cost allocation logic to work with the new Microsoft.FileShares resource type.

Azure Migrate: File Share Assessment Now Available for SMB and NFS

Azure Migrate now discovers and assesses SMB and NFS file shares on Windows and Linux servers, available globally from June 2026. The assessment maps usage, dependencies, and size before migration, and identifies redundant or low-use shares where storage consolidation could reduce cost.

For FinOps teams, the low-use share identification is the most directly actionable output. Large organizations running hundreds of legacy file shares on Windows File Server often have a significant percentage of shares with minimal access. The Azure Migrate assessment quantifies this before migration, making the case for storage reduction rather than lift-and-shift of everything.

For teams evaluating Azure storage commitment strategy alongside the file migration, the Usage.ai guide on Azure Cost Optimization covers Azure storage pricing tiers, commitment options, and right-sizing frameworks.

Azure NetApp Files Migration Assistant

An ANF (Azure NetApp Files) migration assistant launched in June 2026, simplifying migrations from on-premises NAS systems and other cloud storage to Azure NetApp Files. The assistant handles assessment, discovery, and migration job automation. Teams planning ANF migrations should use the assistant to model transfer volumes and timing before committing to ANF capacity and service level tiers. Verify at learn.microsoft.com/azure/azure-netapp-files – rates change.

Azure Databricks: Genie Moves to Pay-As-You-Go on July 6, New Cost Controls Available Now

Databricks Genie: Pay-As-You-Go Billing Starts July 6, 2026

Azure Databricks Genie will move to a pay-as-you-go pricing model on July 6, 2026. Per the official Databricks June 2026 release notes (learn.microsoft.com/azure/databricks/release-notes/product/2026/june – confirmed live):

• Each user receives 150 DBUs of free LLM usage every month, equivalent to approximately $10.50 in the US East region.
• Usage beyond the free allowance is billed in DBUs.

Before July 6, account admins should create budgets in Databricks to monitor and control Genie spending. Budgets can be scoped to Genie at the account level, workspace level, user group level, or individual user level, with email alerts and per-user spending limits configurable. Act before July 6: any Genie usage above the 150 DBU free tier will begin billing on that date.

Cost modeling: if your organization has 100 active Genie users, the total free tier covers approximately 15,000 DBUs per month (roughly $1,050 in US East). Usage patterns that exceed 150 DBUs per user per month should be modeled against your Databricks commitment rates before July 6.

Databricks: statement_text Redaction Starting June 22, 2026

Starting June 22, 2026, the statement_text column in Databricks system.query.history returns Redacted for users who are not account admins or members of the databricks_pii_access account-level group. This is a security change to prevent non-admin users from reading full SQL query text in system tables.

Action required before the rollout completes: audit any workloads, dashboards, or automation that reads statement_text from system.query.history. If the service principal or user running those workloads is not in the databricks_pii_access group or an account admin role, they will start receiving Redacted values after the rollout. Update access grants or move the workflows to a principal with appropriate access before the change takes full effect across your regions.

FinOps and Cost Management: What June 2026 Means for Azure Spend

July 1 RI Deadline: This Is the Last Window

Everything covered in April and May about the Azure Reserved VM Instance retirement lands today. The three-step checklist for any team that has not yet acted:

• Step 1 – Portal check: Go to Azure portal, navigate to Reservations, filter by Virtual Machines, and identify any RI with a VM family in the affected list and an expiry date on or after July 1.
• Step 2 – Choose path: Trade in for Azure Savings Plan for Compute (submit via portal before midnight UTC June 30), purchase a new 1-year RI on the same affected series for a final term (also by midnight UTC June 30), or accept that VMs revert to pay-as-you-go when the RI expires.
• Step 3 – Verify the trade-in or purchase completed: Check Reservations status in the portal immediately after submission to confirm the transaction processed before the cutoff.

Do not assume the trade-in process is instant. Submit with at least two hours of buffer before the midnight UTC cutoff.

Azure HorizonDB and Database Commitment Strategy

Azure HorizonDB public preview introduces a new PostgreSQL-compatible database tier that has different pricing from Azure Database for PostgreSQL Flexible Server. Teams that are planning database commitment purchases should not assume that existing Azure Database for PostgreSQL Savings Plans or Reserved Instances cover HorizonDB instances during preview.

The full guide to Azure Database Savings Plans coverage across Azure SQL, PostgreSQL, Cosmos DB, and other engines – including which services qualify and when Savings Plans outperform Reserved Instances – is on Usage.ai: Azure Database Savings Plans.

Azure Linux: Model the OS License Savings Before Migrating

Azure Linux at no licensing cost creates a cost reduction opportunity for any team running paid Linux distributions on Azure VMs. The migration analysis:

• Identify all Azure VMs running RHEL or SUSE with an OS license cost component on the Azure bill.
• Verify that Azure Linux supports the workload type and OS compatibility requirements for each VM.
• Model the monthly OS licensing cost eliminated per VM.
• Factor in migration testing and any application compatibility work.

For large fleets with hundreds of RHEL or SUSE VMs, the aggregate OS licensing removal can be material. This is separate from the Azure Hybrid Benefit calculation for Windows Server, which applies through a different mechanism.

Databricks Genie Billing: Act Before July 6

The Databricks Genie pay-as-you-go transition on July 6 is an underestimated cost change. Teams that enabled Genie for broad user access without usage controls will start incurring DBU charges per user beyond 150 DBUs per month. Create per-user and per-group spending limits in Databricks Cost Management before July 6.

Configure email alerts at the 80% threshold of the free allowance to catch usage before it bills. Any user consistently above 150 DBUs per month should be identified now and the business case for their usage evaluated before billing starts.