> For the complete documentation index, see [llms.txt](https://docs.archera.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.archera.ai/help-center/cloud-service-intelligence/cloud-service-intelligence/azure/advanced-services/machine-learning.md). # Machine Learning ## Azure Machine Learning — commitment brief > **Data sourced**: March 2026. Verify current figures at the [Azure Pricing Calculator](https://azure.microsoft.com/en-us/pricing/calculator/). ### Coverage summary Azure Machine Learning compute instances run on Azure VMs and receive Reserved VM Instance discounts automatically — no manual assignment required. Both 1-year and 3-year Reserved VM Instances and the Azure Compute Savings Plan apply to AML compute costs, with discounts up to \~72% (3-year RI) and \~65% (3-year Compute Savings Plan) on eligible VM series. Azure Hybrid Benefit also applies to Windows Server or Linux subscription licenses on the underlying VMs. When both a reservation and savings plan exist for the same resource, the reservation discount applies first. **What is covered**: VM compute costs for Azure Machine Learning compute instances and compute clusters, across all eligible VM families (D-series, E-series, N-series GPU, F-series, etc.). **What is not covered**: Windows Server OS license costs (offset with AHB), storage (OS disk — a P10 disk is attached per compute instance and accrues cost even when the instance is stopped), networking, load balancer charges per compute instance (\~$0.33/day), Azure Container Registry, Blob Storage, Key Vault, Application Insights, Azure Monitor, and Private Link/Firewall charges if using managed VNet isolation. ### Azure Hybrid Benefit > AHB applies. Existing Windows Server licenses with Software Assurance remove the Windows OS license cost from the underlying VM bill. For GPU compute instances running Windows, this can add significant additional savings on top of the reservation discount. Linux-based compute instances incur no Windows OS licensing cost. ### SKU / tier coverage | SKU / Tier | Reservation | Savings Plan | 1-yr discount | 3-yr discount | Notes | | ------------------------------- | ----------- | ------------ | ------------- | ------------- | ------------------------------------------------------ | | D/Ds/Dv2–Dv5 (General Purpose) | ✅ Yes | ✅ Yes | \~36–40% | \~55–65% | Most common AML compute instance family | | E/Es/Ev3–Ev5 (Memory Optimized) | ✅ Yes | ✅ Yes | \~36–40% | \~55–65% | | | N-series NC/ND/NV (GPU) | ✅ Yes | ✅ Yes | \~36–40% | \~55–65% | GPU compute; software/driver licenses extra | | F/Fs/Fsv2 (Compute Optimized) | ✅ Yes | ✅ Yes | \~36–40% | \~55–65% | | | M-series (Memory Extreme) | ✅ Yes | ✅ Yes | \~36–40% | up to \~72% | | | A-series (legacy) | ❌ No | ❌ No | — | — | Excluded from VM reservations | | G-series | ❌ No | ❌ No | — | — | Excluded from VM reservations | | Low-priority compute | ❌ No | ❌ No | — | — | Spot-equivalent pricing; no reservation eligibility | | Preview / Promo VMs | ❌ No | ❌ No | — | — | Not eligible while in preview or on promotional meters | ### Reservation vs. Compute Savings Plan **Reservation**: Scoped to a specific VM family, size group, and region. Delivers the highest discount (up to \~72% at 3-yr for some series). Instance size flexibility applies within the same series group by default. Best for stable compute instance families and regions. **Compute Savings Plan**: Hourly spend commitment flexible across VM families, regions, App Service (Pv3/Iv2), Dedicated Host, Container Instances, and Azure Functions Premium. Up to \~65% at 3-yr. Best when compute instance types or regions may change over time. ## Regional availability Azure Machine Learning compute commitments are made at the VM level (Reserved VM Instances or Compute Savings Plan), so availability follows standard VM reservation availability. Reserved VM Instances are available in all commercial Azure regions for eligible VM series. The Azure Compute Savings Plan applies globally across all commercial regions. GPU VM series (N-series) availability is more limited and varies by region — verify N-series availability in your target region before purchasing. **Notable exclusions**: A-series and G-series VMs are excluded from VM reservations in all regions. Low-priority (spot-equivalent) compute is not eligible for reservations in any region. Azure Government regions have limited VM reservation and savings plan availability. ⚠️ Availability varies by SKU and region. Always verify in the [Azure portal](https://portal.azure.com) or via the [Azure Retail Pricing API](https://prices.azure.com/api/retail/prices) before purchasing. ### Archera Azure Machine Learning compute is within Archera's commitment management scope via VM-level reservations and Compute Savings Plans. Archera can automate reservation purchase for AML VM families, monitor utilization, and wrap commitments in a GRI/GSP. *** **Sources** * [Plan to manage costs for Azure Machine Learning](https://learn.microsoft.com/en-us/azure/machine-learning/concept-plan-manage-cost) * [Prepay for Azure Virtual Machines (Reserved VM Instances)](https://learn.microsoft.com/en-us/azure/virtual-machines/prepay-reserved-vm-instances) * [What are Azure savings plans for compute?](https://learn.microsoft.com/en-us/azure/cost-management-billing/savings-plan/savings-plan-overview) ⚠️ Discount percentages vary by VM series, region, and term. Always verify with the [Azure Pricing Calculator](https://azure.microsoft.com/en-us/pricing/calculator/). --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.archera.ai/help-center/cloud-service-intelligence/cloud-service-intelligence/azure/advanced-services/machine-learning.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.