Microsoft Fabric F-SKU Capacity Planning (2026)
F2-F2048 sizing methodology, workload multipliers, PAYG vs RI economics, right-sizing playbook. Fixed-fee 2-week engagement from EPC Group.
Last updated July 7, 2026 by Errin O'Connor, Founder & Chief AI Architect, EPC Group
Microsoft Fabric F-SKU sizing depends on 4 inputs: active concurrent users, OneLake data volume, workload type (BI vs mixed vs heavy-ETL vs real-time), and refresh + concurrency profile. F64 ($5,257/mo PAYG or ~$3,150/mo RI) is the Premium P1 equivalent inflection point. PAYG-to-RI break-even is 60% utilization / 15 hours/day. Migration from Premium P-SKUs: 4-8 weeks for enterprise portfolios. Hidden costs: OneLake storage ($23/TB after 1TB free), Copilot CU consumption (20-30% overhead), autoscale premium rates, connector fees, upskilling. EPC Group 2-week fixed-fee sizing engagement: $25K.
Key Facts
- F64 ($5,257/mo PAYG or ~$3,150/mo RI) is the Premium P1 equivalent inflection
- Workload multipliers: BI 1.0x, Mixed 1.3x, Heavy-ETL 1.8x, Real-time 2.2x
- RI vs PAYG break-even: 60% utilization / 15 hours/day
- Direct SKU equivalence: P1=F64, P2=F128, P3=F256
- Enterprise Fabric TCO: $150K-$500K/year all-in (capacity + storage + services + training)
- EPC Group 2-week fixed-fee capacity sizing engagement: $25,000
Fabric F-SKU Capacity FAQ
What is a Microsoft Fabric F-SKU?
Fabric F-SKUs are capacity units that power all Fabric workloads (Data Warehouse, Data Engineering, Data Science, Real-Time Intelligence, Data Factory, Power BI Premium, and OneLake). Available tiers: F2 (2 CU), F4, F8, F16, F32, F64, F128, F256, F512, F1024, F2048. Each doubles capacity for roughly double the price. F64 ($5,257/month PAYG) is the inflection point — it maps to legacy Power BI Premium P1, includes Copilot in Fabric, and includes Power BI Pro-equivalent access for all viewers.
How do I choose the right F-SKU size?
F-SKU sizing depends on four inputs: (1) Active concurrent users — sum of Power BI dashboard viewers, Fabric notebook developers, and pipeline consumers active at peak. (2) Data volume in OneLake — raw + transformed + gold-zone total. (3) Workload type — BI-heavy (report consumption) vs mixed (BI + Spark ETL) vs heavy-ETL (constant pipelines) vs real-time (event streams). (4) Refresh + concurrency profile — how often datasets refresh and how many concurrent operations at peak. Use the calculator at /tools/microsoft-fabric-capacity-calculator to model your specific case.
What is the workload multiplier?
Different workloads consume different amounts of capacity per user + per GB. EPC Group benchmark multipliers: BI reporting = 1.0x baseline, Mixed BI + ETL = 1.3x, Heavy-ETL (Spark-first) = 1.8x, Real-Time Intelligence + event streams = 2.2x. Apply the multiplier to your user count AND data volume to compute effective capacity requirement. Example: 2,000 concurrent BI users on mixed workload = 2,000 × 1.3 = 2,600 effective users → F32 or F64 depending on data volume.
PAYG vs Reserved Instance — which is right?
PAYG (Pay-As-You-Go) charges per-minute at the F-SKU rate. RI (Reserved Instance) commits to 1 year of capacity for approximately 40% discount. Break-even is around 60% utilization — if you plan to run the capacity more than 15 hours/day, RI wins. EPC Group recommendation: start with PAYG for the first 30-60 days to validate sizing (real workload often differs from projections), then convert to RI once steady-state is established. Example: F64 PAYG = $5,257/month; F64 RI = ~$3,150/month. Annual savings: $25,284 per capacity.
How do I migrate from Power BI Premium P1/P2/P3 to Fabric?
Direct SKU equivalence: P1 = F64, P2 = F128, P3 = F256. Migration path: (1) Purchase new F-SKU in Fabric admin center. (2) In Fabric portal, move each Premium workspace to the new capacity (workspace admin → License mode → change to Fabric). (3) Verify Power BI datasets still refresh + reports still render. (4) Cancel Premium P-SKU. (5) Migrate to Direct Lake mode for 3-10x query performance on large datasets. Zero downtime possible if done during off-peak. EPC Group has completed dozens of P-to-F migrations; typical timeline: 4-8 weeks for enterprise portfolios (100+ workspaces).
What are hidden Fabric capacity costs?
Six hidden costs beyond the F-SKU sticker price: (1) OneLake storage — first 1TB per capacity is free, then $23/TB/month. (2) Copilot in Fabric consumes CU — heavy Copilot use can eat 20-30% of capacity. (3) Autoscale — F-SKU can auto-scale to 2x capacity for peak periods but bills at PAYG rate during scaling. (4) Data egress — cross-region + cross-cloud data movement. (5) Third-party connectors — some ISV connectors bill separately. (6) Training + change management — enterprise Fabric adoption typically needs $50K-$150K of upskilling investment. Full TCO for a mid-size enterprise Fabric deployment: $150K-$500K annually across capacity + storage + services + training.
How does EPC Group help with Fabric capacity planning?
EPC Group Fabric Capacity Sizing engagement is 2-week fixed-fee $25K: (1) Current-state audit — Power BI Premium usage, existing DW workloads, Snowflake/Databricks costs if applicable. (2) Target-state modeling — 3-year forecast of user growth + data volume + workload evolution. (3) F-SKU recommendation with confidence intervals — right-sized to steady-state with autoscale headroom for peaks. (4) PAYG-to-RI transition plan — when to switch based on utilization threshold. (5) ROI model — savings vs current Premium spend, or vs Snowflake/Databricks alternative. Deliverable: Board-ready capacity planning report + fixed-fee proposal for Fabric Foundation Deployment.
Get a Fabric Capacity Sizing
2-week fixed-fee $25K engagement delivers board-ready capacity planning + fixed-fee proposal for Fabric Foundation Deployment. Call (888) 381-9725.
Phone
(888) 381-9725Business Hours
Monday-Friday, 8 AM - 7 PM CT
Quick Response Guarantee
We respond to all inquiries within one business day