Microsoft Fabric F2–F2048 pricing table: Pay-as-you-go + pausing vs 1-year reservation break-even
Choosing the wrong billing model for a Microsoft Fabric capacity is one of the fastest ways to waste 35–50% of your data budget. Leave an F64 on pay-as-you-go (PAYG) running 24/7, and you pay ~$8,410/month when a 1-year Azure reservation costs ~$4,995/month—a $40,980 annual overpayment. Conversely, buy a 1-year reservation for a development or batch ETL capacity that only runs 8 hours on weekdays, and you pay for 550 idle hours every month that you could have paused on PAYG.
This guide gives you the complete F2 to F2048 Microsoft Fabric pricing table at standard PAYG rates ($0.18/CU-hour; pricing varies by agreement and currency), the exact 14.2 hours/day break-even formula between PAYG + pausing and 1-year reservations, and how to blend reserved baseline CUs with burst PAYG capacities. Model your own numbers in our interactive browser savings calculator or download our free Excel Fabric Capacity Savings Checker.
Complete Microsoft Fabric F-SKU pricing table (F2 to F2048): PAYG vs 1-Year Reserved
Microsoft Fabric compute is billed by Capacity Units (CUs). At standard public USD list rates (730 hours/month; pricing varies by agreement and currency; OneLake storage and networking bill separately), each CU costs $0.18/hour on PAYG ($131.40/CU/month) and ~$0.1069/hour on a 1-year reservation (~$78.05/CU/month, a ~40.6% discount):
| Fabric SKU | Capacity Units (CUs) | P-SKU Equivalent | PAYG Hourly | PAYG Monthly (730h) | 1-Yr Reserved Monthly (~41% off) | Annual Saving with 1-Yr RI (if 24/7) |
|---|---|---|---|---|---|---|
| F2 | 2 CU | — | $0.36 / hr | $262.80 / mo | ~$156.10 / mo | ~$1,280 / yr |
| F4 | 4 CU | — | $0.72 / hr | $525.60 / mo | ~$312.20 / mo | ~$2,561 / yr |
| F8 | 8 CU | EM1 / A1 | $1.44 / hr | $1,051.20 / mo | ~$624.38 / mo | ~$5,122 / yr |
| F16 | 16 CU | EM2 / A2 | $2.88 / hr | $2,102.40 / mo | ~$1,248.75 / mo | ~$10,244 / yr |
| F32 | 32 CU | EM3 / A3 | $5.76 / hr | $4,204.80 / mo | ~$2,497.50 / mo | ~$20,488 / yr |
| F64* | 64 CU | P1 / A4 | $11.52 / hr | $8,409.60 / mo | ~$4,995.00 / mo | ~$40,975 / yr |
| F128* | 128 CU | P2 / A5 | $23.04 / hr | $16,819.20 / mo | ~$9,990.00 / mo | ~$81,950 / yr |
| F256* | 256 CU | P3 / A6 | $46.08 / hr | $33,638.40 / mo | ~$19,980.00 / mo | ~$163,901 / yr |
| F512* | 512 CU | P4 / A7 | $92.16 / hr | $67,276.80 / mo | ~$39,960.00 / mo | ~$327,802 / yr |
| F1024* | 1,024 CU | P5 / A8 | $184.32 / hr | $134,553.60 / mo | ~$79,920.00 / mo | ~$655,603 / yr |
| F2048* | 2,048 CU | — | $368.64 / hr | $269,107.20 / mo | ~$159,840.00 / mo | ~$1,311,206 / yr |
The 14.2-hour break-even rule: PAYG + Pausing vs 1-Year Reservation
A 1-year Azure reservation for Microsoft Fabric applies its ~40.6% discount on a strict use-it-or-lose-it hourly basis (Microsoft Learn: Fabric capacity reservations). Reserved CUs that sit paused during an hour are still billed and do not roll over.
Break-Even Active Ratio = 1 − Reservation Discount % = 1 − 0.406 = 59.4% of total hours
• Every day (7 days/week): 24 hours × 0.594 = 14.25 hours/day
• Weekdays only (5 days/week, paused all weekend): 168 hours/week × 0.594 = 99.8 hours/week (~19.9 hours/weekday if paused 48 hours every weekend)
Crucial caveat—carryforward settlement on pause: The raw formula assumes zero unamortized background smoothing debt when you pause. As explained in our pause smoothing and carryforward billing guide, Azure bills remaining 24-hour smoothed background CUs immediately upon pause. Once you factor in typical 15–25% background smoothing tails on production workloads, the real-world break-even for pausing a shared capacity is 10–11 active hours per weekday (~50–55 hours/week).
How single-CU reservations work (blending Reserved + PAYG burst)
In Azure, you do not buy Fabric reservations locked to a rigid SKU name like "F64"—you buy reservations in 1-CU increments. This unlocks two high-ROI FinOps architectures highlighted in our sample 48h check report:
Pattern 1: Reserve the 24/7 baseline, burst on PAYG
Suppose your capacity runs comfortably at F32 (32 CUs) for 20 hours/day, but needs F64 (64 CUs) for 4 hours on weekday mornings (06:00–10:00) during heavy ETL and morning report opens:
- Buy a 1-year reservation for 32 CUs (~$2,498/month).
- Scale the capacity from F32 to F64 for 4 hours on weekday mornings (~88 hours/month).
- During those 4 hours, your 32 reserved CUs cover the first 32 CUs automatically, and Azure bills only the extra 32 CUs at PAYG:
32 CUs × $0.18 × 88 hrs = $506.88/month. - Total cost: ~$3,005/month—saving ~$1,990/month vs a 24/7 reserved F64 ($4,995/mo) and ~$5,405/month vs a 24/7 PAYG F64 ($8,410/mo). (Note: if unlicensed Power BI viewers rely on F64 free viewing, the capacity must stay at F64 while viewers access reports.)
Pattern 2: Split steady BI (Reserved) from spiky Dataflow/Spark ETL (PAYG Paused)
If your capacity suffers from CapacityLimitExceeded throttling or heavy Dataflow Gen2 CU consumption, keep semantic models on a reserved F16 or F32 running 24/7, and run batch pipelines on a separate F8 or F16 PAYG capacity that wakes for 3 hours a day and pauses when done.
Decision matrix: Which billing model should you pick?
| Workload Profile | Active Schedule | Recommended Billing Model | Why |
|---|---|---|---|
| Enterprise BI Hub (200+ free viewers) | 24/7 or 16+ hrs/day | F64 1-Yr Reserved | F64 is mandatory for free viewer licensing, and >14.2 hrs/day beats PAYG pausing. |
| Internal BI (All users have Power BI Pro / E5) | 12–24 hrs/day | F16 or F32 1-Yr Reserved | No F64 license lock-in; reserving a right-sized SKU saves ~41% with zero pause management. |
| Dev / Test / Sandbox Workspaces | 6–9 hrs/day Mon–Fri (~160h/mo) | F2, F4, or F8 PAYG + Automated Pause | Running 160 of 730 hours on PAYG cuts compute cost by ~78%—nearly double the 41% RI discount. |
| Dedicated Nightly Batch ETL | 2–4 hrs/day (~90h/mo) | F8 or F16 PAYG + Job-Triggered Pause | Waking the capacity only for pipeline execution and using native Spark/Copy pays for <13% of the month. |
Frequently asked questions
How much does Microsoft Fabric F64 cost per month?
At standard PAYG rates ($0.18/CU-hour for 730 hours/month; pricing varies by agreement and currency), a Microsoft Fabric F64 capacity (64 CUs) costs $8,409.60/month on pay-as-you-go ($11.52/hour) or approximately $4,995/month with a 1-year Azure reservation (~41% discount). OneLake storage and networking bill separately.
How many hours a day can I run a PAYG Fabric capacity before a 1-year reservation is cheaper?
Because a 1-year reservation discounts compute by ~40.6%, the raw break-even is 14.25 hours/day (7 days/week) or 99.8 hours/week (~20 hours/weekday if paused all weekend). Factoring in 24-hour background smoothing debt billed at pause time, a 1-year reservation is usually cheaper for any shared capacity running more than 10–11 hours per weekday.
If I buy a 32-CU Fabric reservation and scale my capacity to F64 for a few hours, does my reservation still apply?
Yes. Fabric capacity reservations are purchased and applied in 1-CU increments per hour. If you hold 32 reserved CUs and scale a capacity to F64 (64 CUs) for 4 hours, your reservation covers the first 32 CUs and you only pay the PAYG rate ($0.18/CU-hour) for the extra 32 CUs.
Get a free 48h Fabric capacity check
Send a screenshot or CSV export of your Fabric Capacity Metrics app (Compute page, last 14 days) to [email protected]. Within 48 hours you get a one-page diagnostic with your target SKU, pause schedule, carryforward risk, and monthly saving. No tenant access, no call, no cost.
Run the free savings estimator → Download the Excel checker (.xlsx) →