Unexpected High Cost for Dataflows Standard Compute Despite Using Reserved F32 Capacity

[Parthiban006]

Hi community,

I'm using Microsoft Fabric on a Pay-As-You-Go model and have provisioned an F32 capacity for 250 hours/month, with an hourly rate of ₹550. This should result in a total cost of around ₹1.37 lakh.

However, in my Azure billing breakdown, I see a separate meter for "Dataflows Standard Compute Capacity Usage CU" showing a cost of ₹2.6 lakh, which is significantly higher than expected.

Here are some key details:

• No auto-scale is enabled
• All workloads are scheduled within the 250-hour window
• I'm using Dataflow Gen2 for incremental refresh and frequency is every one hour in the buisness hours 9 to 6.
• The meter appears under Fabric Capacity, not as a separate service

Can someone help me understand:

1. Why is this meter showing such a high cost?
2. Is Dataflow Gen2 compute not covered under the reserved F32 capacity?
3. Could this be a billing attribution issue or something else?

Any insights or similar experiences would be greatly appreciated!

Thanks in advance!

[v-karpurapud]

Hi @Parthiban006 

Just checking in as we haven't received a response to our previous message. Were you able to review the information above? Let us know if you have any additional questions.

Thank You.

[v-karpurapud]

Hi @Parthiban006 

We have not received a response from you regarding the query and were following up to check if you had the opportunity to review the information provided. Please feel free to contact us if you have any further questions.

Thank You.

[v-karpurapud]

Hi @Parthiban006 

I wanted to check if you’ve had a chance to review the information provided. If you have any further questions, please let us know. Has your issue been resolved? If not, please share more details so we can assist you further.

Thank You.

[Shahid12523]

Dataflow Gen2 compute is not covered by your reserved F32 capacity.
It’s billed separately under “Dataflows Standard Compute Capacity Usage CU,” even if you stay within your 250-hour window. That ₹2.6 lakh cost likely comes from frequent refreshes and complex transformations.
To reduce cost:
- Lower refresh frequency
- Simplify Power Query steps
- Use the Fabric Metrics App to monitor CU usage

[Anonymous]

Hi @Parthiban006,

Thanks for reaching out to the Microsoft fabric community forum and for sharing all the details, this is a good question and I can clarify how the billing works in your case.

The charges you are seeing under “Dataflows Standard Compute Capacity Usage CU” are not part of your reserved F32 Fabric capacity hours. Dataflow Gen2 currently runs on separate Dataflow Standard compute meters, and these are billed independently of the core Fabric capacity. That’s why you are seeing an additional cost beyond your expected F32 spend. To address your specific points:

* Why the high cost?
Every time your Dataflow Gen2 refresh runs, it consumes compute units that are billed against the “Dataflows Standard Compute” meter. Since you have refreshes scheduled every hour across business hours, the cumulative CU usage adds up and results in the higher bill you’re seeing.

* Is it covered under F32 capacity?

At present, Dataflow Gen2 compute is not covered under the Fabric reserved capacity (F-SKU). Even if you provision an F32 for workloads like lakehouses, warehouses, or semantic models, Dataflow Gen2 continues to be metered and charged separately.

* Is it billing attribution? 

This is not a billing error. It’s the expected behavior today, though the cost breakdown can be confusing since the meter still shows up under “Fabric Capacity” in Azure Cost Analysis.

You can review your Dataflow Gen2 refresh schedules and optimize frequency or transformation logic to reduce CU consumption. If you want all workloads consolidated under Fabric capacity, you may want to consider alternative ingestion patterns (for example, using Notebooks or Pipelines in Fabric instead of Dataflows Gen2 for heavy or frequent refresh tasks). I know this can be surprising if you assumed everything runs within F-capacity, but in your case the billing is correct.

If you want to raise a support ticket for a detailed CU usage breakdown for your Dataflows, kindly go through the below link.
How to create a Fabric and Power BI Support ticket - Power BI | Microsoft Learn

If I misunderstand your needs or you still have problems on it, please feel free to let us know.  

Best Regards,
Hammad.

[Parthiban006]

Hi @Anonymous,

Thanks a lot for the detailed explanation it really helped clarify the billing behavior.

I have a follow-up question regarding the timeline and transparency of this pricing model:

• Is this separate metering and billing for Dataflow Gen2 something that has been in place since the beginning of Fabric, or is it a recent change?
• The Azure Pricing Calculator and most official Microsoft documentation still mention only Compute and Storage as the primary cost components for Fabric. If Dataflow Gen2 is billed separately from Fabric capacity, shouldn't this be explicitly mentioned in the pricing documentation or calculator?

Could you kindly share a link to the official Microsoft documentation that clearly states that Dataflow Gen2 compute is billed separately from Fabric capacity (F-SKU)? This would help me and others in the community better plan and optimize workloads.

Thanks again for your support!

[Anonymous · Accepted solution]

Hi @Parthiban006,

Coming to your "Has this separate billing for Dataflows Gen2 always been around, or is it new?" query, Dataflows Gen2 has always been billed separately from reserved Fabric capacity (F-SKUs). It’s not a recent change but it’s simply how the model has worked from the start, even if that wasn’t always obvious in pricing summaries or calculators.

Kindly go through the below links to get a better understanding of how Dataflow is billed.
Pricing for Dataflow Gen2 - Microsoft Fabric | Microsoft Learn
Data Factory Pricing - Microsoft Fabric | Microsoft Learn

Pricing for data pipelines - Microsoft Fabric | Microsoft Learn

Now coming to "Why doesn't the Azure Pricing Calculator or most Microsoft docs explicitly show this?" query, You’re right most documentation focuses on general Compute and Storage for Fabric, without clearly calling out that Dataflow Gen2 incurs its own compute charges. 

I suggest you to go through a official Microsoft source "Fabric Operations” which shows the billing meter used for Dataflow Gen2. Under the Data Factory section, it lists "Dataflow Gen2 Refresh -> billed via “Dataflows Standard Compute Capacity Usage CU”. These operations are categorized under Fabric, but charged separately. Fabric operations - Microsoft Fabric | Microsoft Learn

That wording directly confirms that Dataflow Gen2 refreshes are metered independently of your reserved F-SKU capacity even though they appear alongside Fabric in billing breakdowns.

Best Regards,

Hammad.