Microsoft Fabric: What It Is, Euro Pricing, vs Power BI

Microsoft Fabric is Microsoft's unified data platform: it brings ingestion, a data lakehouse, a data warehouse, real-time analytics and Power BI together on a single storage layer, OneLake, and it is paid for by capacity. In the West Europe Azure region it costs from €171.46 a month (F2 capacity with a one-year reservation) or €288.35 pay-as-you-go, and Spain Central is around 14% cheaper. At its fiscal 2027 first-quarter earnings call, held on October 28, 2026, Microsoft said it now has over 44,000 paid Fabric customers, up more than 50% year over year (a full transcript is available on fool.com). The question CFOs and BI leads keep asking us is always the same: do I need Microsoft Fabric, or is Power BI enough? This guide answers with euro prices taken from the official Azure API for four European regions, the correct break-even calculation and a comparison with Synapse, Databricks and Snowflake.

Microsoft Fabric in 30 seconds (November 2026)

• What it is: a SaaS analytics platform that combines Data Factory, Spark, warehousing, real-time analytics and Power BI on top of OneLake.
• Price in Europe: F2 from €171.46/month reserved or €288.35/month pay-as-you-go (West Europe, excl. VAT); from €148.12 in Spain Central. Plus about €22/TB of OneLake.
• Licensing: below F64, every Power BI viewer needs Pro (€12.10/month); from F64 a Free license is enough.
• Trial: 60 days free, with 4 or 64 CUs depending on the tenant, no Copilot.
• When it doesn't pay off: a single data source, few reports and nobody with a data background on the team.

What is Microsoft Fabric?

Microsoft Fabric is an end-to-end data analytics platform delivered as software as a service (SaaS) that brings ingestion, transformation, storage, real-time analytics, data science and Power BI reporting into one environment. That is how the official Microsoft Learn documentation defines it: all of its workloads run on a shared compute and storage model. One nuance matters to whoever signs the invoice: you do not need an Azure account for the trial or the Free license, but you do need an Azure subscription to buy a paid F capacity, or a CSP partner, according to the capacity purchasing guide.

The underlying idea is simple: instead of standing up and paying separately for a data orchestrator, a data lake, a data warehouse, a Spark engine and a BI tool, you buy one capacity and every piece consumes it. The main workloads are:

Workload	What it does	Who uses it
Data Factory	Ingestion and transformation with 200+ native connectors and Power Query	Data engineers, advanced analysts
Data Engineering	Processing with Apache Spark and notebooks	Data engineers
Data Warehouse	SQL warehouse with separate compute and storage, in the open Delta Lake format	BI and SQL teams
Databases	Transactional database (SQL database in Fabric) and mirroring	Developers
Real-Time Intelligence	Analysis of data in motion: IoT sensors, logs, clickstream	Operations, manufacturing
Data Science	Training and deploying machine learning models	Data scientists
Power BI	Reports, dashboards and semantic models	Business users, leadership
Fabric IQ (preview)	Shared business semantics, data agents and ontologies	Data architects

Status as of November 2026. Fabric IQ is still in preview and Microsoft ships new features every month.

Underneath all of them sit three shared services: OneLake (the single storage layer), Copilot (generative AI assistance in every workload) and a governance layer centralized in the OneLake catalog, which integrates with Microsoft Purview for sensitivity labels, data loss prevention and auditing.

What is OneLake and how does the Fabric lakehouse work?

OneLake is the single, logical data lake for the whole organization in Microsoft Fabric: each tenant has exactly one OneLake, built on Azure Data Lake Storage Gen2, and every workload reads from and writes to it without moving or duplicating data. Microsoft compares it to OneDrive: just as Word saves to the company OneDrive by default, Spark, the data warehouse and Power BI save to OneLake by default.

The data lakehouse is built on top of OneLake. The Fabric documentation describes it as "a collection of files, folders, and tables that act as a database over a data lake". It combines the low storage cost of a data lake with the structure and SQL querying of a data warehouse. Our data warehouse guide compares warehouse, lake and lakehouse in detail.

Three mechanisms explain why this approach cuts integration work:

• Open Delta Lake format. Data is stored in Delta tables on top of Parquet, an open format other tools can read. You are not locked into a proprietary format.
• Shortcuts and mirroring. Shortcuts let you query data that stays in Amazon S3, Google Cloud Storage or Azure Data Lake without copying it. Mirroring replicates databases such as SQL Server, Azure SQL, Oracle, Snowflake, Google BigQuery or SAP into OneLake in near real time. According to the mirroring documentation, replication compute is free and every capacity unit (CU) you buy includes 1 TB of free storage for replicas: an F64 comes with 64 TB.
• Direct Lake. This Power BI connection mode reads OneLake's Delta tables directly with the VertiPaq engine. According to Microsoft Learn, it delivers performance comparable to Import mode without copying the full data volume on every refresh, because a refresh only updates metadata.

In practice, the most common pattern is the medallion architecture, which organizes the lakehouse into three layers:

Layer	What it holds	How data gets there	Who consumes it
Bronze	Raw data, exactly as it leaves the ERP, CRM or files	Data Factory pipelines, mirroring or shortcuts	Data engineering
Silver	Clean, deduplicated data with correct types	Spark notebooks or Power Query dataflows	Engineers and advanced analysts
Gold	Business tables (sales, margins, stock) in a star schema	SQL or Spark transformations	Power BI via Direct Lake, data agents

How much does Microsoft Fabric cost? Pricing and licensing (F2 to F64)

Microsoft Fabric costs from €171.46 a month (F2 with a one-year reservation) or €288.35 a month pay-as-you-go in the West Europe region, plus about €22 per TB of OneLake and Pro licenses for Power BI viewers as long as you stay below F64. You pay for capacity, not per user: you buy an F SKU (F2 to F2048 and above) whose number is its count of capacity units (CUs), and Azure bills per second with a one-minute minimum and no commitment, or with a discounted one-year reservation, according to the Fabric licensing guide.

Microsoft Fabric price table by SKU (West Europe)

We calculated the prices below from the public Azure Retail Prices API, queried in November 2026 for the West Europe region (Netherlands): €0.1975 per CU per hour pay-as-you-go and €1,028.77 per CU per year with a one-year reservation. The official Fabric pricing page uses the same regional rates. Monthly figures assume 730 hours, VAT excluded:

SKU	CUs	Pay-as-you-go (€/month)	1-year reservation (€/month)	Do Power BI viewers need Pro?
F2	2	288.35	171.46	Yes
F4	4	576.70	342.92	Yes
F8	8	1,153.40	685.85	Yes
F16	16	2,306.80	1,371.70	Yes
F32	32	4,613.60	2,743.39	Yes
F64	64	9,227.20	5,486.78	No (a Free license with a viewer role is enough)

Prices verified November 2026 in the official Azure API (West Europe region, EUR).

Does it cost the same in every region?

No: Fabric pricing depends on the Azure region, and the gap inside Europe reaches 16%. Choose the region for latency and data residency first, but know the price before you commit. Same API query, November 2026:

Region	€/CU-hour (pay-as-you-go)	F2 pay-as-you-go (€/month)	F2 reserved (€/month)	F64 reserved (€/month)
Spain Central (Madrid)	0.1706	249.08	148.12	4,739.89
France Central (Paris)	0.1795	262.07	155.90	4,988.85
Italy North (Milan)	0.1885	275.21	163.68	5,237.82
Germany West Central (Frankfurt) and West Europe	0.1975	288.35	171.46	5,486.78

Extra costs: OneLake, licenses and capacity overage

On top of the capacity you need to add:

• OneLake storage: €0.0215 per GB per month on the hot tier in West Europe, roughly €22 a month per terabyte. Cheaper cool and cold tiers exist for historical data.
• Power BI licenses: Pro costs €12.10 per user per month and Premium Per User (PPU) €20.80, billed annually and excluding VAT; we checked the same euro price on Microsoft's pricing pages for Spain, Germany, France and Italy. Pro is included in Microsoft 365 E5 and Office 365 E5.
• Capacity overage: Fabric absorbs spikes with bursting and smoothing (background operations are spread over 24 hours) and, if you still saturate the capacity, it throttles. You can enable overage billing to avoid throttling, but according to the throttling documentation it costs three times the normal rate: €0.5925 per CU per hour in West Europe. It is the main source of variable spend.

One-year reservation or pay-as-you-go with pausing?

The one-year reservation is 40.5% cheaper than continuous pay-as-you-go (€1,028.77 versus €1,730.10 per CU per year). In exchange, pay-as-you-go lets you pause the capacity: according to the pause and resume documentation, pausing settles accumulated usage and stops compute billing. The math is simple: the reservation pays off if the capacity will be running more than 59.5% of the time, about 434 hours a month. An F4 running 12 hours a day on 22 working days (264 hours) costs €208.56 pay-as-you-go, versus €342.92 reserved.

One warning before you schedule pauses: while the capacity is paused, mirroring stops and replica storage is billed as regular OneLake storage, according to the mirroring documentation. If you replicate your ERP, pausing overnight means stale data in the morning until the replica catches up.

Want to know which SKU your company needs before signing anything? Our Power BI and Fabric consulting team sizes the capacity against your real data volumes.

Microsoft Fabric vs Power BI: what is the difference and when do you need each?

Power BI is the visualization and semantic modeling layer; Microsoft Fabric is the complete data platform that includes Power BI as one of its workloads. If your data is already clean in an ERP or a database and you only need reports, Power BI Pro is enough. If you need to integrate sources, keep history, transform at scale or feed data to AI, you need Fabric (or an equivalent platform).

Criterion	Power BI Pro / PPU	Microsoft Fabric (F capacity)
Pricing model	Per user (€12.10 / €20.80)	Per capacity (from €148/month reserved)
Lakehouse, warehouse, notebooks, pipelines	No	Yes
Central storage (OneLake)	No	Yes
Direct Lake on Delta tables	No	Yes
View reports without a Pro license	No	From F64 only
Copilot in Power BI and Fabric	No (requires capacity)	Yes, from F2
Best for	Teams that only consume reports	Companies that need a data platform

Status as of November 2026.

Two licensing nuances that regularly catch companies off guard:

1. PPU is not Fabric. The licensing guide is explicit: Premium Per User unlocks advanced Power BI features but does not provision Fabric capacity, so it cannot create lakehouses, warehouses or notebooks.
2. Below F64, every viewer needs Pro. On F2 through F32, anyone who opens a Power BI report needs a Pro or PPU license. From F64 up, users with a Free license and a viewer role can see them. Microsoft is also retiring Power BI Premium capacities (P SKUs) and advises new and existing customers to buy F SKUs instead.

Break-even point: when does F64 beat Pro licenses?

F64 pays off when the Pro licenses you save on viewers exceed what it costs on top of the capacity you already need for compute. The right formula is not the full F64 price divided by the Pro price, because the alternative also pays for capacity:

Break-even viewers = (F64 cost − cost of the SKU you need) ÷ €12.10

With West Europe reserved prices:

Capacity you need for compute	Incremental cost of moving to F64 (€/month)	Break-even viewers
F2 (€171.46)	5,315.32	≈ 439
F8 (€685.85)	4,800.93	≈ 397
F16 (€1,371.70)	4,115.08	≈ 340
F32 (€2,743.39)	2,743.39	≈ 227

If your workloads already require an F32, F64 pays for itself from about 227 viewers, not 450. Count only people who read: anyone who creates and publishes Power BI content needs Pro regardless.

How does Microsoft Fabric compare with Synapse, Databricks and Snowflake?

Microsoft Fabric is a capacity-priced SaaS that integrates Power BI and runs only on Azure; Databricks and Snowflake are multicloud, consumption-priced platforms, and Azure Synapse is Microsoft's earlier PaaS service, whose natural successor is Fabric. The choice depends less on technology than on where your ecosystem already lives and how you prefer to pay.

Criterion	Microsoft Fabric	Azure Synapse Analytics	Databricks	Snowflake
Model	SaaS	PaaS	Managed platform in your cloud	SaaS
Pricing	Capacity (CUs), reserved or pay-as-you-go	Dedicated pools (DWU) or serverless per TB processed	Consumption (DBU) plus cloud infrastructure	Compute credits plus storage
Clouds	Azure (reads S3 and GCS via shortcuts)	Azure	AWS, Azure and Google Cloud	AWS, Azure and Google Cloud
Table format	Delta Lake in OneLake	Proprietary in dedicated pools; Delta and Parquet in Spark	Delta Lake (Iceberg via UniForm)	Proprietary and Iceberg
Built-in BI	Native Power BI, Direct Lake	Power BI connector	Own dashboards and connectors	Connectors to BI tools
Best fit	Microsoft 365 and Power BI shops	Existing Azure estates	Engineering- and ML-heavy teams	Multicloud SQL data warehouse

Status as of November 2026.

They are not mutually exclusive: Fabric can mirror Snowflake and read the Azure Databricks catalog through mirroring without copying the data, according to the mirroring documentation. If you are coming from Azure Synapse, Microsoft offers a Fabric Data Warehouse Migration Assistant that copies metadata and data from Synapse dedicated pools, converts the schema automatically and helps fix the T-SQL that doesn't migrate with Copilot; review data type differences first (for example, datetimeoffset is not supported). If you run Power BI Premium (P SKUs), moving to F SKUs is all but mandatory: a P1 matches an F64 in compute.

When does Microsoft Fabric pay off for a mid-sized company?

Microsoft Fabric pays off when the company's problem is not the reports but the data behind them: several sources that never reconcile, manual Excel consolidation, history that nobody keeps, or AI projects that stall because the data is not ready. If your organization already runs on Microsoft 365 and Power BI, Fabric is the lowest-friction next step.

The scenarios where we see it fit best:

• Consolidating ERP, CRM and spreadsheets into a single governed source, with mirroring from SQL Server or shortcuts to existing storage.
• Manufacturing and logistics with real-time data (sensors, telemetry), where Real-Time Intelligence avoids standing up a separate streaming platform.
• Getting data ready for generative AI: data agents and Copilot need a clean semantic model to answer well.
• Replacing a patchwork of tools (orchestrator, data lake, warehouse and BI bought separately) with a single invoice and a single governance model.

Total cost example for a 40-user company

A typical case at West Europe prices: 40 people who view reports, a reserved F4 and 1 TB in OneLake. The F4 costs €342.92, 40 Pro licenses €484.00 and storage €22.02: about €849 a month excluding VAT. If the F4 only runs during business hours on pay-as-you-go with pausing (264 hours), capacity drops to €208.56 and the total to about €715. The no-platform alternative, 40 Pro licenses on top of an existing Azure SQL database, costs €484 in licenses but offers no lakehouse, no history and no Direct Lake.

What we check before sizing a capacity

Three sizing mistakes keep coming up in the Fabric projects we review, and all three stem from how the platform bills, not from the technology:

1. Sizing by number of users. Capacity is consumed mostly by background workloads (refreshes, pipelines, notebooks), which Fabric spreads over 24 hours. That is why we measure with the Capacity Metrics app before proposing a SKU.
2. Reserving before you have a month of usage data. A reservation only pays off above 59.5% uptime; for an intermittent pilot, pay-as-you-go with pausing is cheaper.
3. Pausing without checking replicas. Pausing stops mirroring and bills its storage; a production environment with a mirrored ERP can rarely be switched off overnight.

"In Fabric, the bill isn't driven by the number of users but by what runs in the background: a badly scheduled refresh is spread over 24 hours and can starve the morning reports of capacity. That's why we measure for a month with Capacity Metrics before we talk about reservations." — Alfons Marques, founder of Technova Partners

The available economic evidence is promising but deserves a careful read. The Forrester Total Economic Impact study, commissioned by Microsoft and published in May 2024 (with a summary on Microsoft's blog), estimated a 379% ROI over three years for a composite organization with $5 billion in revenue and 10,000 employees, built from four interviews, with a 25% gain in data engineering productivity and a 20% gain for business analysts. It is a large-enterprise model funded by the vendor: useful for seeing where the savings come from, not as a promise for a mid-sized business.

When it does not pay off yet: if you have a single data source, a handful of reports and nobody on the team with a data background, Power BI Pro and a well-built semantic model will give you a better return. Our data analytics guide for SMBs helps structure that first step, our business intelligence guide explains which dashboards to prioritize, and our comparison of business intelligence tools puts Power BI side by side with its alternatives.

How to get started with Microsoft Fabric without blowing the budget

The safest way to start with Microsoft Fabric is a free trial scoped to one use case, followed by a small pay-as-you-go capacity that is paused outside business hours, and only then a one-year reservation. This is the order we recommend:

1. Activate the 60-day free trial. According to the trial documentation, the trial capacity is configured "as either an F4 capacity (4 capacity units) or an F64 capacity (64 capacity units)" depending on tenant eligibility; if it starts at 4 CUs, the admin can try to raise it to 64 CUs from Capacities. It includes every workload and up to 1 TB in OneLake, but not Copilot or data agents, and it is not meant for production.
2. Pick a use case with a business owner. For example, the sales dashboard someone rebuilds by hand every Monday. A pilot without a business question ends up as an empty lakehouse.
3. Choose your region from day one. Moving a workspace with Fabric items to another region later means deleting them. All EU regions in the price table sit inside the EU Data Boundary, so Copilot processes data without it leaving the EU, according to the Copilot in Fabric documentation.
4. Move to an F2 or F4 on pay-as-you-go and automate pausing in development environments with Azure runbooks. In production with mirroring, weigh the replica warning first.
5. Measure consumption with the Capacity Metrics app for at least a month. Copilot and Direct Lake consume capacity; if you saturate it, Fabric throttles, and overage billing costs three times the rate.
6. Reserve once consumption is stable and above 59.5% of the time. That is when the 40.5% reservation saving outweighs the lost flexibility.
7. Train the team that will run it. Microsoft offers two dedicated certifications: DP-600 (Fabric Analytics Engineer) and DP-700 (Fabric Data Engineer). Our corporate AI training covers using Copilot and data agents on Fabric.

Keep the limits of the small SKUs in mind: on F2 to F8, a Direct Lake model can use at most 10 GB and 3 GB of memory, according to the Direct Lake limits table. For most mid-sized companies that is plenty; for a group processing millions of transactions a day, it is not.

If you would rather not go it alone, our Power BI and Fabric implementation service covers everything from the pilot to data governance, and our data strategy service helps you decide up front which platform fits your roadmap.

Evaluating Microsoft Fabric, or torn between Fabric and Power BI Pro? Tell us about your case and we will prepare a capacity and cost estimate based on your real data.

Microsoft Fabric in 2026: what to buy for your situation

The decision comes down to four profiles:

• You only need reports on data that is already clean: Power BI Pro at €12.10 per user, no Fabric.
• You want to test Fabric on one use case: the 60-day trial, then an F2 or F4 on pay-as-you-go with pausing in development.
• You run a data platform in production with stable usage: a one-year reservation of the SKU Capacity Metrics points to, in the cheapest region that meets your residency needs.
• You have many viewers: use the incremental break-even formula; with an F32 baseline, F64 pays off from about 227 viewers.

For context on Microsoft's AI ecosystem, read our Microsoft Copilot 365 business guide and explore our data analytics and business intelligence services.

Frequently Asked Questions

Is Microsoft Fabric free?

Not entirely. The Fabric (Free) license costs €0 and is assigned when you first sign in, but without a capacity it cannot run anything. To try it there is a 60-day free trial with 4 or 64 CUs and 1 TB in OneLake, without Copilot. After that you need a paid F SKU, from €148.12 a month reserved in Spain Central or €171.46 in West Europe.

How much does Microsoft Fabric cost per month?

In West Europe, an F2 costs €288.35 a month pay-as-you-go or €171.46 with a one-year reservation, excluding VAT (official Azure API, November 2026). OneLake is billed separately, about €22 per TB a month, and below F64 every Power BI viewer needs a Pro license (€12.10 a month).

What does the Microsoft Fabric F64 license include?

F64 provides 64 capacity units, matches the former Power BI Premium P1 in compute and is the first SKU where viewers with a Free license and a viewer role can see Power BI reports without Pro. It includes 64 TB of free mirroring storage. It costs €5,486.78 a month reserved in West Europe.

What is a CU and which SKU do I need?

A CU (capacity unit) is Fabric's measure of compute power: an F2 has 2 CUs and an F64 has 64. The right SKU depends on the volume of refreshes, pipelines and queries, not on the number of users. For a mid-sized company F2 to F8 is usually enough; confirm it with a month of data in the Capacity Metrics app.

Do I need an Azure subscription to use Microsoft Fabric?

Not for the free trial or the Free license. To buy a paid F capacity, yes: it is purchased through an Azure subscription linked to your Microsoft Entra tenant, or through a CSP partner who manages that subscription for you. Billing arrives alongside your other Azure services.

Does Microsoft Fabric replace Azure Synapse?

For most new projects, Fabric is the destination Microsoft proposes: its Data Warehouse offers a Migration Assistant from Synapse dedicated pools that copies metadata and data and converts the schema automatically. Synapse keeps working, but it is worth planning the migration and reviewing unsupported data types such as datetimeoffset.

Which certification exists for Microsoft Fabric?

Microsoft offers two associate-level certifications: DP-600, Fabric Analytics Engineer, focused on modeling, Power BI and analytics; and DP-700, Fabric Data Engineer, focused on ingestion, Spark and real-time data. You prepare with free Microsoft Learn paths and renew them each year through an online assessment.