Microsoft Fabric replaces Power BI Premium and introduces a new capacity licensing model. Understanding F-SKUs, OneLake cost structures, and how Fabric fits your EA is essential before you commit.
Microsoft Fabric is Microsoft's unified data platform, launched in 2023 as the strategic successor to a fragmented set of Azure data services including Azure Data Factory, Azure Synapse Analytics, Azure Data Explorer, and Power BI Premium. Rather than managing separate services with separate pricing, Fabric consolidates data integration, data engineering, data warehousing, real-time analytics, data science, and business intelligence into a single platform on top of a unified storage layer called OneLake.
The commercial significance of Fabric is substantial. For organisations with existing Power BI Premium investments, Fabric represents both an upgrade path and a pricing model change. For organisations building new data platforms, Fabric is the clear Microsoft strategic direction — and understanding its licensing model before committing is essential to avoiding unexpected costs. For those evaluating Microsoft's broader cloud investment in the context of their Microsoft Enterprise Agreement, Fabric is increasingly a meaningful line item alongside Azure consumption and M365.
Before Fabric, organisations managed separate licences for Power BI Premium (for BI), Azure Synapse (for data engineering), Azure Data Factory (for pipelines), and Azure Data Explorer (for real-time analytics). Fabric unifies all of these under a single capacity-based licence. The trade-off: simpler architecture, but a new licensing model to understand — and new cost structures that don't always favour existing customers.
Microsoft Fabric is licensed on a capacity model based on Fabric Capacity Units (CUs). Each F-SKU provides a fixed number of CUs shared across all Fabric workloads running on that capacity. Larger F-SKUs provide more CUs, supporting heavier workloads, more concurrent users, and faster processing at proportionally lower unit cost.
Want independent help negotiating better terms? We rank the top advisory firms across 14 vendor categories — free matching, no commitment.
| F-SKU | Capacity Units (CUs) | Approx. Monthly Price | Approx. Annual Price | Best For |
|---|---|---|---|---|
| F2 | 2 CU | ~$263 | ~$3,156 | Development, testing, small teams |
| F4 | 4 CU | ~$526 | ~$6,312 | Small BI workloads |
| F8 | 8 CU | ~$1,052 | ~$12,624 | Dept-level analytics |
| F16 | 16 CU | ~$2,104 | ~$25,248 | Mid-size BI + data engineering |
| F32 | 32 CU | ~$4,208 | ~$50,496 | Enterprise BI + Spark workloads |
| F64 | 64 CU | ~$8,416 | ~$100,992 | Large enterprise analytics |
| F128 | 128 CU | ~$16,832 | ~$201,984 | High-scale data platform |
| F256 | 256 CU | ~$33,664 | ~$403,968 | Enterprise-wide analytics hub |
| F512 | 512 CU | ~$67,328 | ~$807,936 | Large-scale data + AI workloads |
| F1024+ | 1024+ CU | $134,656+ | $1.6M+ | Hyperscale analytics |
The unit economics of F-SKUs improve substantially at higher tiers. An F64 provides 32x the capacity of an F2 at approximately 32x the price — so scaling up is roughly linear in cost-to-capacity terms. However, F-SKU sizing requires careful workload analysis: over-provisioning wastes spend, while under-provisioning causes throttling and poor user experience on report loads and data refreshes.
The most common Microsoft Fabric licensing mistake is sizing based on peak concurrent workloads rather than average utilisation. Fabric capacities smooth demand with built-in burst and background processing queues — many organisations can achieve 80% of peak performance needs with a capacity one tier below what their peak analysis suggests. Right-sizing with a load simulation before committing to annual capacity saves significant annual spend.
For existing Power BI Premium P-SKU customers, Microsoft Fabric presents a migration decision: stay on P-SKUs (which are being deprecated) or migrate to equivalent F-SKUs. The mapping is not one-to-one in all cases, and the commercial implications vary by organisation.
| Power BI Premium P-SKU | Equivalent F-SKU | Price Change | Capability Delta |
|---|---|---|---|
| P1 (8 vCores) | F64 | ~$8,416/mo vs ~$4,995/mo P1 | +Full Fabric workloads (Spark, Warehouse, Real-Time) |
| P2 (16 vCores) | F128 | ~$16,832/mo vs ~$9,990/mo P2 | +Full Fabric workloads |
| P3 (32 vCores) | F256 | ~$33,664/mo vs ~$19,980/mo P3 | +Full Fabric workloads |
| P4 (64 vCores) | F512 | ~$67,328/mo vs ~$39,960/mo P4 | +Full Fabric workloads |
The Fabric migration from P-SKUs to F-SKUs represents a 60–70% price increase for equivalent capacity. Microsoft justifies this premium through the additional workload capabilities included in Fabric — data engineering, real-time analytics, and data warehousing that would previously have required separate Azure service spend. If your organisation was already spending on Azure Synapse, Data Factory, or Data Explorer alongside Power BI Premium, the consolidated Fabric cost may be broadly comparable or even lower when all Azure data services are accounted for. If you were running Power BI Premium exclusively for BI, the Fabric migration is a clear cost increase that warrants negotiation.
Fabric's capacity model handles compute, but users still need appropriate individual licences to create, publish, and consume Fabric content. Understanding the per-user licence requirements alongside the capacity licence is essential to calculating total Fabric TCO.
Get the IT Negotiation Playbook — free
Used by 4,200+ IT directors and procurement leads. Oracle, Microsoft, SAP, Cloud — all covered.
| User Role | Licence Required | Included In | Standalone Cost |
|---|---|---|---|
| Report viewer (read-only, Fabric capacity workspace) | None (with capacity) | Covered by F-SKU capacity | $0 |
| Report creator / analyst | Power BI Pro | M365 E3, E5, some bundles | $10/user/mo standalone |
| Data engineer / notebook author | Fabric (Free) or Pro | Free tier for Fabric workloads | Free (limited) |
| Premium content publisher | Power BI Pro or PPU | M365 E3+; Pro standalone | $10/mo Pro or $20/mo PPU |
| Power BI Premium Per User (isolated) | PPU add-on | Not included in standard M365 | $20/user/mo |
The licensing model for viewers is one of Fabric's genuine advantages over the old Power BI model. Under Power BI Premium capacity, external users viewing reports in capacity-backed workspaces do not need Power BI Pro licences — the capacity licence covers viewing access. For organisations sharing analytics broadly with large internal user populations or external stakeholders, this can dramatically reduce per-user licence cost compared to assigning Pro licences to every report consumer. For organisations with M365 E3, Power BI Pro is already included — minimising incremental user licence cost for Fabric deployments. See our M365 E3 vs E5 comparison for bundle context.
Fabric capacity can be purchased in two commercial models: Azure pay-as-you-go (billed per hour with pause/resume capability) or pre-committed through Microsoft EA or Azure Reserved Capacity. The right model depends on your workload pattern and risk appetite.
| Factor | Azure PAYG | EA / Azure Reserved Capacity |
|---|---|---|
| Flexibility | High — pause, scale, cancel | Low — committed for 1–3 years |
| Price per CU | Full list price | 17–40% discount (1yr/3yr) |
| Pause to save cost | Yes — stop billing when idle | No — committed regardless of use |
| Predictable budgeting | Variable — depends on usage | Fixed monthly commitment |
| MACC contribution | Only consumed amount | Full committed amount counts to MACC |
| Best for | Batch/scheduled workloads, dev/test, early adoption | Production BI, always-on analytics, mature workloads |
The optimal approach for most organisations is a hybrid: PAYG for development environments and batch workloads that can be paused, Reserved Capacity for production BI workloads that must always be available. This balances cost efficiency for variable workloads against pricing certainty for production analytics that has predictable, continuous demand.
Ready to optimise your Microsoft Fabric licensing?
Microsoft Fabric is increasingly negotiable within Enterprise Agreement structures, particularly for organisations making significant data platform commitments. The key principles for Fabric EA negotiation follow the same logic as broader Microsoft EA negotiation — commitment, competitive alternatives, and strategic value create the conditions for better pricing.
Commit to a data platform migration narrative: Microsoft is highly motivated to win Fabric workloads that currently sit on AWS (Redshift, Glue, S3), Google (BigQuery, Dataproc), or Snowflake/Databricks. If your organisation is evaluating Fabric against these alternatives, make that competition explicit. Microsoft will invest in protecting platform migrations it considers strategically significant — and Fabric vs Snowflake or Fabric vs Databricks is precisely that kind of conversation.
Bundle Fabric with Azure MACC: Fabric capacity can be funded through Azure consumption commitments under MACC. Including Fabric in your Azure MACC negotiation — essentially committing to Fabric capacity as part of your total Azure spend pledge — can improve your MACC discount tier across all Azure services while securing favourable Fabric capacity pricing. See our Azure committed spend guide for MACC tier optimisation context.
Negotiate P-to-F migration credits: For Power BI Premium customers migrating to Fabric F-SKUs, there is room to negotiate migration credits that offset the price increase inherent in moving from P-SKUs to equivalent F-SKUs. Microsoft has offered transition assistance programmes for P-SKU customers, but these are typically not proactively disclosed — you need to request them explicitly as part of the migration conversation. Engaging a specialist Microsoft negotiation advisor who has navigated P-to-F migrations with other clients is the most reliable route to accessing these programmes.
Use Databricks and Snowflake as credible alternatives: Both Databricks and Snowflake are mature, enterprise-grade data platforms that compete directly with Microsoft Fabric on data engineering and warehousing workloads. Running parallel proof-of-concepts on both platforms — and presenting the results and total cost comparison to Microsoft — creates genuine competitive pressure that Microsoft responds to with better pricing.
Expert right-sizing, EA negotiation, and P-to-F migration support typically reduces Fabric total cost by 20–35% compared to standard Microsoft proposals.