Power BI Premium to Microsoft Fabric Migration
P-SKUs are retired. This Power BI Premium to Fabric Migration guide covers the verified deadlines, exact SKU equivalents, real cost numbers, step-by-step migration methods, and the mistakes teams make when they rush the cutover.
Non-EA customers: P-SKUs reached end-of-life on January 1, 2026. EA customers: Deadline is January 1, 2028, or when the EA term ends — whichever comes first. After your subscription ends, Microsoft provides 30 days of free equivalent capacity, then throttled data access for 90 days. Total window before data loss risk: approximately 120 days from subscription end. The equivalent F-SKU for a P1 is F64. F-SKUs are purchased through the Azure portal, not the Power BI portal. Pricing varies by region — verify current rates using the Azure Pricing Calculator.
Non-EA customers hit end-of-life on January 1, 2026. EA customers have until January 1, 2028, or their EA term end — whichever is earlier. After expiry: 30 days free capacity, then 90 days throttled access, then data suspension. Large tenants need 3–6 months to migrate safely.
Executive Summary – Power BI Premium to Fabric Migration
A quick reference for IT leadership and architecture planning teams.
| Question | Answer |
|---|---|
| P1 equivalent | F64 |
| P2 equivalent | F128 |
| Non-EA deadline | Jan 1 2026 |
| EA deadline | Jan 1 2028 |
| Free grace period | 30 days |
| Data access window | Up to 90 days |
| Purchase location | Azure Portal |
Power BI Premium & P-SKU Retirement Timelines
Every date that matters, sourced from the official Microsoft Power BI blog and Microsoft Learn FAQ.
New customers could no longer buy Power BI Premium P-SKUs from this date.
P-SKU renewals ended for all customers. Every renewal from this date must be a Fabric F-SKU purchased through Azure.
Non-Enterprise Agreement customers reached end of life. The grace window began: 30 days free capacity, then throttling, then suspension.
Enterprise Agreement customers must migrate before their EA term ends, or by this hard date — whichever comes first. Sovereign cloud customers (GCC, GCC-High, DoD, China) are exempt — Fabric is not available there.
- Days 1–30: Free Power BI Premium capacity matching your previous P-SKU size.
- Days 31–90: Throttling kicks in — interactive operations are heavily throttled and delayed.
- Day 90+: Capacity may be frozen; access to Power BI data could be lost. Capacities left in this state are eventually removed by Microsoft.
GCC, GCC-High, DoD, and China cloud customers are unaffected — Microsoft Fabric is not available in those environments. Power BI Premium remains fully supported there. Confirm your cloud environment type before planning anything.
Tenants that waited for renewal to force the migration are the ones that had problems. Testing RLS, embedded tokens, semantic models, and scheduled refresh on the new F-SKU takes time you won’t have under deadline pressure. Six to eight weeks of parallel running is the realistic minimum for a mid-size tenant.
P-SKU to F-SKU Mapping
Each P-SKU maps to an equivalent F-SKU in compute capacity. The difference is how you purchase them, the billing flexibility, and what workloads they unlock beyond Power BI.
| P-SKU (Retiring) | F-SKU Equivalent | vCores | Cost (Approx US East PAYG) | Typical Use Case |
|---|---|---|---|---|
| P1 | F64 | 8 vCores | ~$5,300/mo* | Department / Mid-size BI |
| P2 | F128 | 16 vCores | ~$10,600/mo* | Enterprise BI |
| P3 | F256 | 32 vCores | ~$21,200/mo* | Large Enterprise |
| P4 | F512 | 64 vCores | ~$42,400/mo* | Multi-region deployment |
| P5 | F1024 | 128 vCores | ~$84,800/mo* | Very large scale Fabric workloads |
Every Power BI Premium feature works identically on the equivalent F-SKU: paginated reports, XMLA endpoints, deployment pipelines, large dataset storage, row-level security, and scheduled refresh. F-SKUs additionally unlock Direct Lake, Copilot, Lakehouse, Data Warehouse (see our structural assessment framework comparing Microsoft Fabric vs Snowflake capabilities), Real-Time Intelligence, and Data Factory.
At F64 and larger, Power BI report consumers do not need a Pro licence. Below F64, every viewer still needs Pro ($14/user/month). For organisations with more than ~350 report viewers, F64 is typically cheaper than per-user licensing alone (350 users × $14 = $4,900/month, roughly matching F64 paused costs).
F-SKUs are purchased through Azure, not the Power BI admin portal. In enterprise environments this means procurement needs to raise an Azure subscription request — which can add 2–4 weeks of internal approval time. Bring procurement into the conversation at the same time you start capacity planning — not after.
Prerequisites & Pre-Migration Checklist
Skipping the assessment phase is where migrations break down. Before reassigning a single workspace, get a full picture of what you’re moving and where it’s going.
- Inventory all P-SKU capacities and workspacesList every Premium capacity — P-SKU tier, Azure region, assigned workspaces, and workspace admins. Export from Admin Portal → Capacities. This spreadsheet becomes your migration tracker.
- Run 90–120 day usage reportsUse the Power BI capacity metrics app and Azure Monitor to understand peak refresh times, dataset sizes, and query concurrency. This helps determine whether the equivalent F-SKU is correctly sized. (See our Fabric capacity optimization guide).
- Flag all Fabric-native workspacesIdentify workspaces containing Lakehouse, Warehouse, Dataflow Gen2, Pipelines, or Notebooks. These need additional steps — especially for cross-region moves.
- Confirm your Azure region strategySame-region moves are seamless. Cross-region moves for Fabric-native workspaces require export, delete, migrate, and recreate. Know this before starting.
- Notify workspace owners 2–3 weeks aheadActive jobs cancel when a workspace is reassigned. They restart automatically, but users see errors if they’re mid-refresh at cutover time. Communication prevents most support tickets.
- Provision F-SKU capacity and validate before migratingYour new F-SKU must be active and tested before reassigning any workspace. Confirm the correct Azure region for data residency compliance and that admin permissions are configured.
Migration Methods
Three approaches exist. The right one depends on the number of workspaces and whether Git integration is in use.
Method 1 — Manual (Workspace by Workspace)
Navigate to each workspace → Settings → Premium → assign to the new Fabric capacity. Best for pilots, small environments, or testing before committing to bulk moves.
Method 2 — Admin Portal Bulk Migration
In the Power BI Admin Portal, select multiple workspaces and reassign them in one operation. Scheduled refreshes pause during the move and resume automatically. The right approach for mid-size tenants migrating 10–50 workspaces by department.
Method 3 — PowerShell and REST API Automation
Microsoft provides official bulk migration tooling alongside REST API and PowerShell scripts. For tenants with 50+ workspaces, this is the only practical option.
- Detach the Git repository before migrating the workspace
- Complete the workspace reassignment to the new F-SKU capacity
- Re-attach the repository and sync all artifacts from the source branch
- Skipping the detach step causes sync conflicts that are slow to untangle
- Export or script-backup all Lakehouse, Warehouse, Dataflow Gen2, Pipelines, and Notebook content first
- Delete these Fabric items from the original workspace before the region move
- Migrate the workspace to the new capacity in the target region
- Recreate or restore Fabric items in the new region after migration completes
On a large enterprise migration spanning three Azure regions, the most time-consuming part was post-migration validation, not the migration itself. We built a test suite covering RLS, scheduled refresh, embedded tokens, and XMLA connections for every workspace before sign-off. Build your validation checklist before you move anything.
Fabric Capacity Pricing & Cost Decision: PAYG vs Reserved
Microsoft Fabric cost models are completely different from old Premium tiers. F-SKUs are billed by the hour through Azure. The PAYG rate looks higher than a P-SKU monthly price at first glance — but F-SKUs can be paused, which completely changes the real-world economics.
Pay-As-You-Go (PAYG)
- No commitment, pause any time
- Pause nights and weekends — cuts cost 55–65%
- Best for variable or seasonal workloads
- Start here, collect 60 days of data before committing
Reserved Capacity (1–3 Year)
- Up to ~40% discount vs PAYG 24/7 rate
- Billed even when capacity is paused
- Best for steady-state always-on workloads
- Only reserve after 60 days of real usage data
F64 vs P1 Comparison Matrix
| Scenario | Monthly Cost | Annual Cost | Notes |
|---|---|---|---|
| P1 PAYG | ~$4,995 | ~$59,940 | Always-on. Cannot pause. Power BI only — no Fabric workloads. |
| F64 PAYG — paused nights & weekends | ~$2,900 | ~$34,800 | ~12hr/day weekdays (~260hrs/month). All 6 Fabric workloads included. |
| F64 Reserved — 1 year | ~$5,000 | ~$60,000 | ~40% off PAYG 24/7. Billed even when paused. Best for always-on workloads. |
| F64 PAYG — 24/7 | ~$8,400 | ~$100,800 | Worst case — no pause strategy. Only justified for true 24/7 demand. |
The single most costly mistake after migration: forgetting to delete the old P-SKU capacity. Microsoft does not cancel P-SKU billing when you reassign workspaces. Delete the old capacity in the Admin Portal immediately after cutover — set a calendar reminder 48 hours out.
Common Migration Issues & Fixes
Permissions Errors After Workspace Reassignment
RLS rules, workspace roles, and embedded tokens tied to the old capacity can break. Run audit logs and access tests before and after. If embedded tokens fail, regenerate them bound to the new F-SKU capacity ID — the old ID in token generation calls is the most common cause.
Active Jobs Cancelled at Cutover
Reassigning a workspace cancels jobs running at that exact moment. They restart automatically on the new capacity. Schedule migrations during off-peak hours. Scheduled jobs not currently running are unaffected.
API Rate Limit Failures on Large Tenants
Scripting bulk migrations across 100+ workspaces hits API rate limits. Migrate in batches of 20–30, log workspace IDs on failure, and build retry logic. A single-pass full-tenant script consistently fails at scale.
Capacity Provisioned in the Wrong Azure Region
Your F-SKU must be in the same Azure region as your existing data for data residency compliance. Procurement often defaults to their nearest region. Confirm the region before the Azure resource is created — changing it after means deprovision and reprovision.
Only Power BI Premium per capacity P-SKUs are being retired. Power BI Pro and Premium Per User (PPU) licences are separate products and completely unaffected. Pro and PPU users access content on F-SKU capacities without any licence changes.
Migration Validation Checklist
After migrating to Fabric capacities, verify these components across all reassigned workspaces to prevent end-user disruption:
- Scheduled RefreshEnsure dataset and dataflow refreshes trigger successfully on the new capacity.
- XMLA EndpointsTest read/write operations from external tools like SSMS or Tabular Editor.
- Row-Level Security (RLS)Verify security roles continue to apply correctly to end users.
- AppsCheck that published Power BI Apps load and distribute content properly.
- Embedded ReportsConfirm tokens are generated against the new F-SKU capacity ID.
- Deployment PipelinesValidate that dev/test/prod stages can still promote content seamlessly.
- DataflowsEnsure upstream dataflows refresh without timeout errors.
- Semantic ModelsCheck query performance of complex DAX statements.
- Gateway ConnectionsConfirm on-premises data gateways connect smoothly to the F-SKU datasets.
Frequently Asked Questions
Pricing figures and deadlines are based on Microsoft’s official documentation and Azure pricing pages at time of writing (June 2026). Always verify current rates at azure.microsoft.com/pricing/details/microsoft-fabric and confirm deadlines with your Microsoft account representative before any purchasing decision. UIG Data Lab is an independent publication, not affiliated with or endorsed by Microsoft Corporation.
