Power BI & Microsoft Fabric for Regulated Enterprises
By Errin O'Connor, Chief AI Architect & CEO of EPC Group | Updated April 2026
A comprehensive guide for healthcare, banking, and government organizations navigating the Power BI to Fabric transition. Where Power BI ends and Fabric begins, licensing implications, governance requirements, migration timing, and capacity planning — built from EPC Group's regulated-industry deployment experience.
The Core Question: Do You Need Fabric or Just Power BI?
Microsoft's messaging blurs the line between Power BI and Fabric, creating confusion for enterprise buyers. Here is the distinction:
- Power BI is a business intelligence tool: semantic models, reports, dashboards, paginated reports, deployment pipelines. It consumes data that has already been prepared.
- Microsoft Fabric is a unified analytics platform that includes Power BI plus data engineering (lakehouses, notebooks), data integration (Data Factory), real-time analytics (Eventhouse), and data science workloads. Fabric produces and governs the data that Power BI consumes.
If your only need is reporting and dashboarding on data that is already clean and accessible, Power BI alone is sufficient. Fabric adds value when you need to unify data engineering, real-time analytics, and BI under one governance umbrella — which is increasingly the case for regulated enterprises managing data across multiple source systems.
Where Power BI Ends and Fabric Begins
| Capability | Power BI (Standalone) | Requires Fabric |
|---|---|---|
| Interactive dashboards & reports | Yes | No |
| Paginated reports | Yes (Premium) | No |
| Semantic models (datasets) | Yes | No |
| Dataflows Gen2 | Limited | Full capability |
| Lakehouse / Delta tables | No | Yes |
| Spark notebooks | No | Yes |
| Data Factory pipelines | No | Yes |
| Real-time analytics (Eventhouse) | No | Yes |
| OneLake (unified storage) | No | Yes |
| Shortcuts to external data | No | Yes |
| Copilot in Fabric | No | Yes (F64+) |
Licensing: P-SKUs vs F-SKUs
Microsoft is retiring Power BI Premium P-SKUs in favor of Fabric F-SKUs. Here is what regulated enterprises need to know:
| SKU | Capacity Units | Approx. Monthly Cost | Notes |
|---|---|---|---|
| F2 | 2 CU | ~$260 | Dev/test only |
| F8 | 8 CU | ~$1,040 | Replaces EM/A1 |
| F16 | 16 CU | ~$2,080 | Small production |
| F32 | 32 CU | ~$4,160 | Mid-size production |
| F64 | 64 CU | ~$8,320 | Replaces P1. Min for Copilot, Trusted Workspace |
| F128 | 128 CU | ~$16,640 | Replaces P2. Large enterprise |
| F256+ | 256+ CU | ~$33,280+ | Replaces P3+. Multi-workload enterprise |
Key licensing considerations for regulated enterprises: Power BI Pro licenses ($10/user/month) are still required for individual users unless they access content through Fabric capacity with Power BI per-capacity licensing. F64 is the minimum for Trusted Workspace and Copilot features. Autoscale allows burst capacity but must be budget-capped to avoid cost surprises.
Governance Architecture for Regulated Industries
Governance is where regulated enterprises cannot cut corners. Here is the governance stack we deploy:
- Microsoft Purview Integration: Purview sensitivity labels applied to Fabric lakehouses, semantic models, and reports. PHI/PII data automatically classified and protected.
- Data Access Governance: Row-level security (RLS) for report consumers. Object-level security (OLS) to hide sensitive columns. Workspace-level permissions mapped to Azure AD security groups.
- Lineage and Impact Analysis: Fabric's lineage view tracks data flow from source to lakehouse to semantic model to report. Essential for regulatory audit trails.
- Endorsement and Certification: Only "Certified" datasets can be used for regulatory reporting. Prevents shadow analytics from untrusted data sources.
- Audit Logging: Unified audit log captures every action: who accessed what, when, from where. Retention policies configured to meet regulatory requirements (7 years for financial services, 6 years for HIPAA).
- Data Loss Prevention: DLP policies prevent download or export of datasets containing PHI/PII/financial data to unapproved destinations.
Decision Tree by Industry
Healthcare (HIPAA)
- Current state: Power BI Premium connecting to Epic Caboodle or Cerner HealtheAnalytics via gateway.
- Fabric value: OneLake shortcuts to EHR data warehouses eliminate ETL pipelines. Real-time analytics for ED operational dashboards. Unified governance for PHI across all analytics workloads.
- Migration timing: Plan transition in 2026 when P-SKU retirement timelines are finalized. Start with F64 pilot on non-PHI workloads.
- Caution: Verify that your BAA covers Fabric-specific services, not just Power BI. Some Fabric workloads (data science notebooks) may require additional compliance review.
Banking and Financial Services (SOC 2, SOX)
- Current state: Power BI Premium with heavy use of paginated reports for regulatory filing, DirectQuery to on-premises SQL Server data warehouses.
- Fabric value: Data Factory replaces SSIS/ADF for ETL orchestration. Lakehouse architecture consolidates siloed data marts. Real-time analytics for fraud detection and trading dashboards.
- Migration timing: Financial services firms should pilot Fabric with non-SOX workloads first. SOX-scoped reports require validated migration with audit committee approval.
- Caution: Gateway-dependent workloads (on-premises SQL Server) still need the gateway in Fabric. Fabric does not eliminate gateway requirements for on-premises data.
Government (FedRAMP, IL4/IL5)
- Current state: Power BI Premium GCC or GCC High, often with Air Force, Army, VA, or civilian agency-specific compliance requirements.
- Fabric value: Unified analytics in GCC reduces the number of separate tools agencies must authorize. OneLake provides a governed data lake within the GCC boundary.
- Migration timing: Wait for full feature parity in GCC/GCC High (currently lagging commercial by 3-6 months). Pilot with non-classified workloads in GCC.
- Caution: Fabric in GCC High and IL5 environments has limited workload availability. Verify specific workloads (notebooks, real-time analytics) are authorized in your environment before planning migration.
Capacity Planning for Regulated Workloads
Right-sizing Fabric capacity is critical for budget management. EPC Group's capacity planning methodology:
- Baseline current usage: Use the Power BI Premium Capacity Metrics app to measure current CU consumption over 30 days. Identify peak hours and top-consuming workloads.
- Map to F-SKU: Current P1 customers typically need F64 or F128. Do not assume 1:1 mapping; Fabric workloads (notebooks, data factory) add CU consumption on top of Power BI workloads.
- Plan for Fabric workloads: If you will run data engineering (Spark notebooks, lakehouse) on the same capacity, add 30-50% CU headroom above your Power BI baseline.
- Separate dev/test: Use F2 or F4 for development. Never run development workloads on production capacity in regulated environments.
- Configure autoscale with budget caps: Fabric autoscale can burst capacity during peak loads but must be budget-capped (e.g., 2x base capacity maximum) to prevent cost overruns.
- Monitor monthly: Set up capacity alerting at 70% and 90% utilization. Review monthly with the analytics governance committee.
Migration Roadmap: Power BI Premium to Fabric
EPC Group's phased migration approach for regulated enterprises:
- Phase 1 (Weeks 1-4): Assessment. Inventory all Power BI workspaces, datasets, reports, gateways, and data sources. Classify by regulatory sensitivity. Identify Fabric-eligible workloads.
- Phase 2 (Weeks 5-8): Capacity Migration. Convert P-SKU to F-SKU. Reassign workspaces. Validate that all existing Power BI functionality works on Fabric capacity without regression.
- Phase 3 (Weeks 9-16): Workload Migration. Migrate selected ETL pipelines to Fabric Data Factory. Create lakehouses for consolidated data sources. Build OneLake shortcuts to replace gateway-dependent connections where applicable.
- Phase 4 (Weeks 17-20): Governance and Testing. Configure Purview integration, sensitivity labels, DLP policies. Validate RLS and OLS across migrated workloads. Conduct compliance review with internal audit.
- Phase 5 (Weeks 21-24): Production Cutover. Switch production workloads to Fabric. Decommission legacy P-SKU. Monitor capacity utilization for 30 days. Optimize and right-size.
Frequently Asked Questions
Do we need Microsoft Fabric if we already have Power BI Premium?
Not necessarily. If your analytics needs are met by Power BI Premium (semantic models, paginated reports, deployment pipelines, and scheduled refreshes), Fabric adds cost and complexity without immediate ROI. Fabric becomes necessary when you need: a unified lakehouse for data engineering, real-time analytics via Eventhouse, data science notebooks, or cross-workload data sharing that eliminates ETL between separate platforms. EPC Group recommends a Fabric readiness assessment before committing to migration.
How does Fabric licensing work for regulated enterprises?
Fabric uses capacity-based licensing (F-SKUs: F2 through F2048), measured in Capacity Units (CUs). Power BI is included in every Fabric SKU. For regulated enterprises, key considerations are: F64 is the minimum for HIPAA workloads (includes Trusted Workspace support), multi-geo requires F-SKUs with geo-specific capacity allocation, and autoscale must be evaluated against budget controls. Power BI Premium P-SKUs are being retired in favor of F-SKUs, so all regulated enterprises should plan their transition timeline.
Is Microsoft Fabric HIPAA-compliant?
Yes, when properly configured. Fabric is covered under Microsoft's BAA and includes HIPAA-eligible services. However, compliance is a shared responsibility. You must: enable Purview sensitivity labels on lakehouses and semantic models, configure private endpoints for network isolation, implement row-level and object-level security, enable audit logging via unified audit log, restrict data export and download capabilities for PHI-containing workspaces, and maintain data residency within US regions. EPC Group has deployed Fabric in HIPAA environments for health systems already using Epic and Cerner data.
What is the realistic migration timeline from Power BI Premium to Fabric?
For most enterprise deployments: 2 to 4 weeks for assessment and planning, 4 to 8 weeks for capacity migration (P-SKU to F-SKU conversion, workspace reassignment), 4 to 12 weeks for workload migration (converting data pipelines to Fabric Data Factory, migrating datasets to lakehouses where beneficial), and 2 to 4 weeks for governance reconfiguration and testing. Total: 3 to 7 months. The key variable is how many existing Power BI workloads should move to Fabric lakehouses versus staying as standalone semantic models.
Should government agencies adopt Fabric or stay on Power BI Premium GCC?
Fabric is now available in GCC and GCC High environments, so the compliance barrier is cleared. However, government agencies should evaluate carefully: GCC Fabric features lag behind commercial by 3 to 6 months, FedRAMP-aligned consulting expertise work for specific Fabric workloads may vary, and many agency data governance policies require review before adopting lakehouse architecture. EPC Group recommends government clients pilot Fabric with non-sensitive workloads in GCC while maintaining Power BI Premium GCC for production reporting. Migrate production workloads only after validating governance controls in the Fabric GCC environment.
Plan Your Fabric Migration with Confidence
EPC Group offers a 2-week Fabric Readiness Assessment for regulated enterprises. We audit your current Power BI environment, classify workloads by regulatory sensitivity, model F-SKU capacity requirements, and deliver a phased migration roadmap. Our team includes Power BI architects with deep experience in HIPAA, SOC 2, and FedRAMP environments.
Call (888) 381-9725 or schedule a consultation below.
Request a Fabric Readiness Assessment