At a glance
| SQL support | T-SQL in Fabric Data Warehouse and the read-only SQL analytics endpoint of each lakehouse; KQL for real-time data |
| Storage and compute architecture | OneLake storage in Delta Lake format, separate from a shared capacity of capacity units |
| Workload types | Data integration, Spark engineering, warehousing, real-time analytics, data science, BI |
| Integrations | Power BI, Microsoft 365, Microsoft Purview, Microsoft Foundry, mirroring and shortcuts to other clouds |
| Deployment | Microsoft-hosted SaaS; F capacities bought through Azure |
| Pricing model | F SKU capacity (F2 to F8192), pay-as-you-go per second or reserved; OneLake storage per GB-month |
Fabric suits organisations standardised on Microsoft 365 and Power BI that want one SaaS capacity for ingestion, lakehouse, warehouse and BI, and teams stronger in T-SQL and low-code tools than in cluster management. Size capacities with care: all workloads share them, heavy jobs can throttle interactive work, and Power BI licensing changes at F64.
Overview
Microsoft Fabric is delivered as software as a service inside a Microsoft Entra tenant. Microsoft lists its workloads as Power BI, Databases, Data Factory, Industry Solutions, Real-Time Intelligence, Data Engineering, Data Science, Data Warehouse and Fabric IQ, which is in preview. All of them store data in OneLake, a single tenant-wide logical data lake built on Azure Data Lake Storage Gen2; Fabric Data Warehouse stores its tables natively in Delta Lake format.
Compute is a capacity that every workload in its workspaces consumes. Shortcuts give zero-copy access to data in ADLS, Amazon S3, Google Cloud Storage and other locations, and mirroring replicates sources such as Azure SQL Database, Azure Cosmos DB, Azure Databricks and Snowflake into OneLake. Microsoft also positions Fabric as the destination for Azure Synapse Analytics workloads.
Key features
- OneLake: one logical lake per tenant on ADLS Gen2, with shortcuts to ADLS, S3 and Google Cloud Storage
- Fabric Data Warehouse: T-SQL warehouse on Delta tables with multi-table transactions
- Data Engineering: Apache Spark notebooks and jobs, with starter pools and custom pools
- Data Factory pipelines and Power Query dataflows with more than 200 native connectors
- Power BI built in, including Direct Lake reports over OneLake tables
- Real-Time Intelligence with eventstreams, Eventhouse and KQL databases
- Mirroring of Azure SQL Database, Cosmos DB, Azure Databricks and Snowflake into OneLake
Pros and cons
Pros
- One capacity covers ingestion, Spark, T-SQL warehouse, real-time analytics and Power BI
- Power BI built in, with Direct Lake mode over OneLake tables
- Shortcuts and mirroring bring in data from other clouds and platforms without copying
- Familiar T-SQL for SQL Server teams, and a 60-day F64-equivalent trial
Cons
- Microsoft cloud only; no deployment on AWS or Google Cloud
- All workloads share the capacity, so overloads trigger throttling of new operations
- Power BI viewers need Pro or PPU licences unless the capacity is F64 or larger
- Iceberg support works through metadata virtualisation, with documented limits
Pricing
Fabric compute is bought as a capacity: F SKUs from F2 (2 capacity units) to F8192, purchased through Azure and billed per second with a one-minute minimum and no commitment, or as a one- or three-year reservation that the Azure pricing page says saves about 41% against pay-as-you-go. Every workload assigned to the capacity draws on it. OneLake storage is billed per GB per month on top. Pricing is regional; the Azure pricing page shows figures only after choosing a region, but Microsoft's Azure Retail Prices API lists pay-as-you-go capacity usage at USD 0.18 per CU-hour and OneLake hot storage at USD 0.026 per GB-month in East US (November 2026), which implies about USD 0.36 per hour for an F2. On capacities smaller than F64, every Power BI viewer also needs a Pro, Premium Per User or individual trial licence. A 60-day free trial gives an F64-equivalent capacity.
Pricing checked on the vendor's official page on 7 November 2026. Prices change; confirm before buying.
Who it is best for
Microsoft 365 and Power BI organisations that want lakehouse, warehouse, pipelines and BI in one SaaS product with a single capacity bill.
Who should look elsewhere
- Teams that need the same platform on AWS or Google Cloud, or deep Spark and ML control: see Databricks vs Microsoft Fabric.
- Teams that only need a managed SQL warehouse across clouds: see Snowflake vs Azure Synapse, which also covers where Fabric fits.
Verdict
Fabric is the simplest route to a combined lakehouse, warehouse and BI stack for Microsoft-centred organisations, and it is where Microsoft now directs new warehousing work. Plan capacity sizing, workload separation and Power BI licensing up front, and compare it with Databricks or Snowflake if you need more than one cloud.
Frequently asked questions
How is Microsoft Fabric priced?
You buy a capacity (F2 to F8192 capacity units) through Azure, billed per second with a one-minute minimum or reserved for one or three years at about 41% less. OneLake storage is billed separately per GB per month. Microsoft's Azure Retail Prices API listed capacity usage at USD 0.18 per CU-hour in East US in November 2026.
Do I need an Azure subscription to use Fabric?
Microsoft states that you do not need an Azure account to use Fabric itself, but F capacities are purchased through Azure. Older Power BI Premium P capacities also support Fabric, and Microsoft is retiring P SKUs in favour of F SKUs.
Is there a free trial of Fabric?
Yes. The Fabric trial lasts 60 days and provides a capacity equivalent to F64.
Sources
- Microsoft Learn: What is Microsoft Fabric?
- Microsoft Learn: Microsoft Fabric licenses and capacity
- Microsoft Learn: Capacity throttling and smoothing
- Microsoft Learn: What is Fabric Data Warehouse?
- Microsoft Fabric pricing
- Azure Retail Prices API (Microsoft Fabric meters)
Checked November 2026.
How we research reviews and what star ratings mean: our editorial method.