Microsoft Fabric. F2 starts at $263. F2048 lands at $269,108. The Capacity Unit math is where the negotiation lives.

Microsoft Fabric is the unified data and AI platform that replaces Power BI Premium per Capacity, Synapse Analytics, and parts of Azure Data Factory under a single commercial vehicle.

The pricing unit is the Capacity Unit (CU), packaged into F SKUs from F2 (2 CUs) at around $263 per month to F2048 (2,048 CUs) at around $270,000 per month. One year reserved capacity reduces the rate by approximately forty one percent against pay as you go.

All workloads draw from the same CU pool. That is the architectural shift that creates both the opportunity and the trap. The covered workloads include:

• Data Engineering
• Data Factory
• Data Warehouse
• Data Science
• Real Time Intelligence
• Power BI
• Databases

Customers can buy a single capacity for everything, but the workload mix matters because Spark, warehouse queries, and Power BI all consume CUs at very different rates. The Power BI Premium per Capacity transition forced most existing Power BI Premium customers into Fabric F64 or larger, frequently at materially higher cost.

This paper sets out actual F SKU pricing, CU consumption math by workload, the reservation discount, the migration path from Power BI Premium, the Snowflake and Databricks competitive frame, and the eleven move buyer side playbook. Read the related Microsoft services practice, the Microsoft knowledge hub, and the Microsoft Azure commitment negotiation article.

What you will learn

• The F SKU ladder and pay as you go pricing. F2 ($263), F4 ($526), F8 ($1,051), F16 ($2,102), F32 ($4,204), F64 ($8,409), F128 ($16,819), F256 ($33,638), F512 ($67,277), F1024 ($134,554), F2048 ($269,108) per month at list. One year reservation cuts about forty one percent off pay as you go.
• How Capacity Units consume across workloads. A Spark notebook consumes CUs at a different rate than a Warehouse query than a Power BI report. The workload mix determines the right F SKU.
• The Power BI Premium per Capacity migration. What changed when Microsoft retired P SKUs, why most P1 customers landed on F64, and the real cost delta.
• The OneLake economics. Storage in OneLake is billed separately from compute capacity, at $24 per TB per month, with cross workload sharing through OneCopy and shortcuts.
• The reservation versus pay as you go decision. When the one year reservation discount makes sense, when the pay as you go pause and resume capability is worth the premium.
• The Snowflake and Databricks competitive frame. Where each competitor wins on functionality, where they win on price, and how to use the alternative credibly.
• Fabric Copilot. The AI capabilities Microsoft layers on Fabric, the additional CU consumption they drive, and the cost impact.
• The eleven move buyer side playbook. Sequenced from workload telemetry through F SKU sizing and competitive process, with dollar values against each move.

Table of contents

Who this is for

CIOs, Chief Data Officers, VPs of IT Procurement, Data Platform leaders, BI Center of Excellence owners, and procurement leaders running Microsoft Fabric at scale. Useful for customers migrating from Power BI Premium per Capacity (P SKUs) to Fabric (F SKUs), customers consolidating Synapse Analytics and Azure Data Factory onto Fabric, and customers evaluating Fabric against Snowflake or Databricks for the next data platform decision.