Power BI P1 to Fabric F64 migration: Do you really need F64, or can you run on F32?

By The CostFabric team · 27 October 2026 · 9 min read

With Power BI Premium per-capacity SKUs (P1, P2, P3) retired in favor of unified Microsoft Fabric capacities, thousands of data teams face the same renewal question: does our Power BI P1 tenant really need to migrate to a Fabric F64, or can we drop to an F32 (or split capacities) and cut the bill in half?

In raw compute, a Power BI Premium P1 (8 v-cores) maps to a Microsoft Fabric F64 (64 Capacity Units). At standard PAYG rates ($0.18/CU-hour, or ~$8,410/month for 730 hours; pricing varies by agreement and currency) or ~$4,995/month on a 1-year reservation, moving 1:1 from P1 to F64 keeps your invoice flat. Yet in our capacity audits—previewed in our sample 48h check report—over 60% of P1 tenants average under 28% CU utilization across the week and only spike above 70% during morning refresh collisions.

Here is the engineering and licensing framework to decide whether to stay at F64, downsize to F32, or split into a hybrid F32 + F8/F16 architecture.

The #1 hard rule: When F64 is mandatory for Power BI viewer licensing

Before looking at Capacity Unit charts, evaluate the F64 licensing boundary (Microsoft Learn: Fabric licenses):

• On F64 or higher (F64, F128, F256+): Users who only view Power BI reports, apps, and paginated reports do not need a paid Power BI Pro or Premium Per User (PPU) license. They only need a Viewer role and a Microsoft Fabric (Free) license. Report creators still need Power BI Pro.
• On F2 through F32: Free viewer distribution is not included. Every user who opens a Power BI report on an F2, F4, F8, F16, or F32 capacity must hold a paid Power BI Pro license (or Microsoft 365 E5, which includes Power BI Pro).

Four P1 migration sizing scenarios compared

Compare the four primary target architectures at standard list pricing (pricing varies by agreement and currency; see our full F2–F2048 PAYG vs Reserved pricing table):

Why P1 capacities look "full" when they only need an F32

Many admins see red spikes above 100% at 08:30 in the Capacity Metrics app on P1 and assume downsizing to F32 is impossible. Three fixable patterns usually explain why a P1 looks saturated:

1. Top-of-the-hour schedule collisions

When dozens of semantic models and dataflows are scheduled at 06:00, 07:00, and 08:00, they compete for memory and interactive threads while stacking 24-hour background smoothing curves. Staggering refreshes onto 15-minute offsets and moving batch refreshes before 05:00 typically cuts peak smoothed utilization by 25–40%.

2. Unoptimized Dataflow Gen2 staging and full refreshes

A few Dataflow Gen2 items with default staging enabled can consume 15–30 CUs of smoothed background capacity all day. Replacing staging dataflows with Copy Activities or incremental refreshes often frees 16 to 32 CUs—the entire gap between F64 and F32.

3. Dev/test notebooks sharing the production P1

On P1, teams put executive dashboards, sandbox notebooks, and UAT pipelines on one capacity because P-SKUs were monolithic. In Fabric, you can provision an F2 or F8 in seconds and pause it outside work hours—protecting production reports from interactive delay and rejection while letting the primary capacity drop a tier.

How to test F32 readiness in 10 minutes from Capacity Metrics

1. Open the Microsoft Fabric Capacity Metrics app on the Compute page for your P1 or F64 over the last 14 days.
2. On the Utilization (CU %) over time chart, click More options (…) → Export data → Summarized data (.xlsx or .csv).
3. Paste the export into our free Excel Fabric Capacity Savings Checker (or run a quick estimate in our browser savings calculator).
4. Check two thresholds:
   • 14-day Average CU % on P1/F64: If average utilization is below 32% of 64 CUs (<20.5 CUs), your average load fits comfortably inside an F32 (~64% average utilization on 32 CUs).
   • Top 5 Items by Total CU (s): If your top 3 background items drive >45% of total CU seconds, optimizing them (disabling Dataflow Gen2 staging, enabling incremental refresh) will bring your P95 below the F32 ceiling.
5. Test on PAYG for 48 hours before reserving: Migrate to an F-SKU on PAYG first, watch carryforward overage when scaling or pausing, confirm zero throttling across two weekly cycles, and only then buy a 1-year reservation.

Frequently asked questions

Is Microsoft Fabric F64 identical to Power BI Premium P1?

In compute capacity, yes: both P1 and F64 provide 64 Capacity Units (8 v-cores equivalent) and unlock free Power BI report viewing for users without a Power BI Pro license. Unlike P1, F64 is an Azure resource that supports per-second pay-as-you-go billing, live scaling up or down, and pausing.

What happens to Power BI report viewers if we downgrade from P1/F64 to F32?

On any capacity smaller than F64 (F2 through F32), free Power BI viewer access is disabled. Every user who views a Power BI report on an F32 capacity must hold a Power BI Pro or Premium Per User (PPU) license (including Pro licenses bundled in Microsoft 365 E5).

Should I buy one F64 or two F32 capacities?

Two F32 capacities cost the same as one F64 (64 CUs total) and isolate production BI from heavy ETL or dev workloads so runaway jobs never throttle executive dashboards. However, two F32 capacities do NOT unlock free Power BI report viewing—a single capacity must be F64 or larger for unlicensed viewers.