Capacity and SKUs in Power BI Embedded
Understanding capacity and SKUs in Power BI Embedded is essential for controlling costs, ensuring performance, and selecting the right licensing model for your embedded analytics solution. The SKU landscape has evolved significantly with the introduction of Microsoft Fabric, and navigating the options requires clarity on workloads, virtual cores, and pricing tiers. At EPC Group, we have architected Power BI Embedded solutions for ISVs, SaaS platforms, and enterprise portals, and we help clients avoid the costly mistake of over-provisioning or under-sizing their capacity.
What Is Power BI Embedded Capacity?
Power BI Embedded capacity is a pool of dedicated compute resources allocated exclusively to your embedded analytics workloads. Unlike Power BI Pro, which runs on shared infrastructure, capacity provides guaranteed resources that are not affected by other customers' workloads. This dedicated allocation ensures predictable performance for your embedded reports and dashboards, which is critical for customer-facing applications where slow load times directly impact user satisfaction.
Capacity is measured in virtual cores (v-cores) and memory. Each SKU tier provides a specific number of v-cores and memory, which determine how many concurrent users your embedded solution can support, how quickly reports render, and how large your datasets can be. The more v-cores you have, the more parallel operations the capacity can handle, including data refreshes, report renders, and DAX queries.
It is important to understand that capacity is a shared resource across all workspaces assigned to it. If you assign ten workspaces to a single capacity, they share the same v-cores and memory. Properly sizing your capacity and distributing workloads across it requires an understanding of your usage patterns, peak concurrency, and dataset sizes.
The SKU Landscape: A, EM, P, and F SKUs
Microsoft has offered multiple SKU families for Power BI capacity over the years. With the introduction of Microsoft Fabric, the SKU naming has been consolidated, but understanding the history helps you navigate documentation and existing deployments:
- A-series SKUs (Azure) - Purchased through Azure and billed hourly. These were the original Power BI Embedded SKUs (A1-A8), designed for ISVs and developers who wanted pay-as-you-go pricing. They can be paused when not in use, making them cost-effective for non-24/7 workloads. A-series SKUs do not include access to the Power BI Service for internal users.
- EM-series SKUs (Microsoft 365) - Smaller capacity SKUs (EM1, EM2, EM3) purchased through Microsoft 365 admin center. These were designed for organizations that needed capacity for embedding but at a lower price point than P-series. EM SKUs include limited Power BI Service features.
- P-series SKUs (Premium) - The full Power BI Premium capacity SKUs (P1-P5), providing the complete Power BI Service experience plus embedding rights. P-series includes all Premium features: paginated reports, deployment pipelines, XMLA endpoints, large datasets, and AI capabilities. These are billed monthly.
- F-series SKUs (Fabric) - The newest capacity SKUs introduced with Microsoft Fabric. F-series (F2-F2048) replace and consolidate the previous SKU families. F64 and above include Power BI Premium features. F-series SKUs provide access to all Fabric workloads (Power BI, Data Factory, Synapse, Real-Time Intelligence) on a single capacity.
SKU Comparison: Choosing the Right Tier
Selecting the right SKU requires balancing cost, performance, and feature requirements. Here is a practical comparison of the most commonly deployed tiers:
| SKU | V-Cores | Memory (GB) | Max Dataset Size | Best For |
|---|---|---|---|---|
| F2 / A1 | 2 | 3 | 3 GB | Dev/test, small apps |
| F8 / A2 | 8 | 10 | 10 GB | Small production apps |
| F64 / P1 | 64 | 25 | 25 GB | Enterprise, full Premium features |
| F128 / P2 | 128 | 50 | 50 GB | Large enterprise, high concurrency |
The critical threshold is F64 (equivalent to P1). Below F64, you get embedding capabilities but not the full Power BI Premium feature set. At F64 and above, you unlock paginated reports, XMLA endpoints, deployment pipelines, large dataset support, AI visuals, and all other Premium features. This threshold often drives the decision between a lower SKU with Power BI Embedded (A-series) and a higher SKU with full Premium capabilities.
Cost Optimization Strategies
Power BI Embedded capacity can be expensive if not managed carefully. Our consultants implement several cost optimization strategies:
- Auto-pause and auto-resume - Azure A-series and F-series SKUs can be paused programmatically when not in use (evenings, weekends) and resumed before business hours. This can reduce costs by 60-70% for non-24/7 workloads. Use Azure Automation or Logic Apps to schedule pause/resume operations.
- Right-sizing - Start with a smaller SKU and monitor utilization through the Power BI Premium Capacity Metrics app. Scale up only when utilization consistently exceeds 70-80% during peak hours. Over-provisioning is the most common cost mistake we see.
- Workspace distribution - Assign workspaces to capacity strategically. Place high-priority, customer-facing reports on dedicated capacity while keeping internal development workspaces on shared (Pro) capacity.
- Dataset optimization - Reduce dataset sizes through proper data modeling (removing unnecessary columns, using integer keys instead of text keys, implementing aggregation tables). Smaller datasets consume less memory and allow more datasets to coexist on a given capacity.
- Refresh scheduling - Stagger data refresh schedules to avoid concurrent refresh storms that spike resource consumption. Use incremental refresh for large datasets to minimize refresh duration and resource usage.
Embedding Scenarios: App Owns Data vs User Owns Data
Power BI Embedded supports two embedding patterns, each with different authentication, licensing, and capacity implications:
App Owns Data (embed for your customers). Your application authenticates using a service principal or master account, and end users do not need Power BI licenses. This pattern is used by ISVs and SaaS providers embedding analytics in their products. Capacity (A-series or F-series SKU) is required. Row-level security is enforced programmatically through effective identity in the embed token.
User Owns Data (embed for your organization). End users authenticate with their own Azure AD credentials, and each user needs a Power BI Pro or PPU license (or Premium/Fabric capacity for free viewers). This pattern is used for internal enterprise portals where reports are embedded in SharePoint, Teams, or custom intranet applications. The user's existing Power BI permissions and RLS assignments apply automatically.
Why Choose EPC Group for Power BI Embedded Architecture
With 28+ years of enterprise Microsoft consulting experience, EPC Group architects Power BI Embedded solutions that are correctly sized, cost-optimized, and production-ready. We have helped ISVs embed analytics in SaaS products serving thousands of tenants, and we have built enterprise portals with embedded Power BI for Fortune 500 clients across healthcare, finance, and manufacturing.
Our approach starts with a capacity assessment that models your expected workload: concurrent users, report complexity, dataset sizes, refresh frequency, and peak usage patterns. We then recommend the optimal SKU, configure auto-scaling policies, and implement monitoring dashboards so you can track utilization and costs in real time. This data-driven approach prevents both over-spending and under-provisioning.
Need Help Sizing Your Power BI Embedded Capacity?
Contact EPC Group for a capacity assessment and SKU recommendation. We model your workload requirements, optimize costs, and deploy a production-ready embedded analytics solution.
Frequently Asked Questions
What is the difference between Power BI Embedded and Power BI Premium?
Power BI Embedded (A-series SKUs) is an Azure resource designed for ISVs and developers embedding analytics in custom applications, billed hourly with pay-as-you-go pricing. Power BI Premium (P-series or F64+ SKUs) provides the full Power BI Service experience including all Premium features, plus embedding rights. Premium is billed monthly. With Microsoft Fabric, the F-series SKUs unify both scenarios: F2-F32 replace A-series for embedding, F64+ replace P-series for full Premium plus embedding.
Can I pause Power BI Embedded capacity to save costs?
Yes, Azure-based A-series and F-series capacities can be paused and resumed through the Azure portal, PowerShell, Azure CLI, or REST API. When paused, you are not charged for compute (only storage charges may apply). This is ideal for dev/test environments and applications that do not require 24/7 availability. Pausing is not available for P-series SKUs purchased through Microsoft 365. EPC Group typically implements automated pause/resume schedules to reduce costs by 60-70%.
How many concurrent users can a specific SKU support?
There is no fixed concurrent user limit per SKU because performance depends on report complexity, dataset size, DAX calculation intensity, and interaction patterns. As a rough guideline, an F8 (A2) can support 50-100 concurrent users with moderately complex reports. An F64 (P1) can support 500-1000+ concurrent users. EPC Group conducts load testing during capacity assessment to validate these estimates against your specific reports and usage patterns.
Do end users need Power BI licenses for embedded reports?
For the "App Owns Data" pattern (embedding for external customers), end users do not need Power BI licenses. The capacity SKU covers all rendering costs. For the "User Owns Data" pattern (embedding for internal users), each user needs a Power BI Pro or PPU license unless the content is in a Premium capacity (F64+/P1+), in which case free Power BI users can view the embedded content.
Should I use Microsoft Fabric SKUs instead of A-series for new projects?
Yes, for new projects, Microsoft Fabric F-series SKUs are the recommended path. They provide a unified capacity that supports Power BI plus all other Fabric workloads (Data Factory, Synapse Data Engineering, Real-Time Intelligence). A-series SKUs remain available but do not include Fabric capabilities. If you only need Power BI Embedded without other Fabric workloads, A-series still works and may be more cost-effective for small deployments, but the long-term strategic direction is F-series.