Microsoft Fabric Capacity Calculator
F-SKU recommendation + monthly cost. Power BI Beta Team heritage.
Workload Profile
Recommendation
Prices are Microsoft list prices (East US, Azure Retail Prices API, read October 28, 2026): $0.18 per CU-hour pay-as-you-go and $938 per CU per year on a 1-year reservation; your region and agreement may differ. The users and data thresholds behind the recommendation are EPC Group planning heuristics, not Microsoft limits; confirm with the Fabric Capacity Metrics app.
How the recommendation is calculated (assumptions version 2026-10-28)
Effective users = peak concurrent users × the workload multiplier; effective data = OneLake GB × the same multiplier. The recommendation is the smallest F-SKU whose user and data thresholds both cover the effective load. Monthly cost = capacity units × $0.18 × 730 hours on pay-as-you-go, or capacity units × $938 ÷ 12 on a 1-year reservation.
Workload multipliers (EPC Group planning heuristics): BI reporting only ×1 · Mixed BI + occasional Spark ×1.3 · Heavy ETL + Spark + ML ×1.8 · Real-Time Intelligence + Data Activator ×2.2
| F-SKU | CU | Effective users up to | Effective GB up to | Pay-as-you-go / month | Reservation / month |
|---|---|---|---|---|---|
| F2 | 2 | 50 | 100 | $263 | $156 |
| F8 | 8 | 200 | 500 | $1,051 | $625 |
| F16 | 16 | 500 | 1,500 | $2,102 | $1,251 |
| F32 | 32 | 1,500 | 5,000 | $4,205 | $2,501 |
| F64 | 64 | 5,000 | 20,000 | $8,410 | $5,003 |
| F128 | 128 | 15,000 | 100,000 | $16,819 | $10,005 |
| F256 | 256 | 50,000 | 500,000 | $33,638 | $20,011 |
| F512 | 512 | no upper threshold | no upper threshold | $67,277 | $40,021 |
The user and data thresholds are planning heuristics, not Microsoft limits; confirm sizing with the Fabric Capacity Metrics app. Download the assumptions (CSV).
Email me this recommendation + get a senior architect review
We'll send the summary to your inbox and a senior EPC Group architect will follow up within the next business day. No spam.
Schedule Your Fabric Capacity Discovery
Power BI Beta Team founding-member heritage. Microsoft Press Power BI author.