TL;DR
Microsoft Fabric pricing is based on capacity, which is a shared pool of computing power. Every Fabric tool, from reports to data pipelines, draws from that pool. The smallest size, F2, costs about $263 a month in most US regions. A larger F64 capacity costs about $8,410 a month. Committing for one or three years cuts those prices by about 41%. Storage and licenses for report builders cost extra, and on F64 or larger, report readers need no paid license.
Key Takeaways Microsoft charges for Fabric as capacity. You rent a pool of capacity units (CUs), and then every workload draws from it. In East US, an F2 costs $262.80 a month on pay-as-you-go and an F64 costs $8,409.60, with a reservation cutting about 41%. A reservation beats pay-as-you-go once a capacity runs more than about 60% of the hours in a month. Below F64, every Power BI report viewer needs a Pro, PPU or trial license. On F64 and above, viewers can use the Free license. Storage, Copilot, capacity overage and Spark autoscale add to the bill, so budget for them alongside the SKU price. Power BI Premium P SKUs are retiring, so P SKU customers need an F SKU plan before their agreement ends. Watch on YouTube
How Microsoft Fabric Helps Reduce Cost and Boost Data Performance
A Kanerika case study walkthrough on how a Fabric implementation lowered data platform costs while speeding up reporting, the same trade-offs this pricing guide covers.
The Budget Meeting Where the Fabric Invoice Gets Questioned Picture a quarterly budget review at a manufacturer that moved its reporting to Microsoft Fabric six months ago. The data team bought an F32 because it looked right for the workload. Finance now asks why the Power BI line went up when the whole point of Fabric was one bill.
The answer sits in two decisions that nobody wrote down. The team sized the capacity for compute, but nobody checked who would open the reports. On an F32, all 300 people reading the sales dashboard need a paid Power BI license. At $4,200 a month, that line costs more than the reserved F32 itself.
Both decisions are cheap to get right before the purchase order and expensive to unwind after a one-year reservation. For that reason, every price below comes from Microsoft’s own price feed, checked in September 2026.
How Microsoft Fabric Pricing Works in 2026 Microsoft Fabric uses a capacity-based pricing model. In practice, you buy a Fabric capacity in Azure, sized as an F SKU, and that capacity supplies a fixed number of capacity units. Every Fabric experience runs on the same pool. This includes Power BI, Data Factory pipelines , Spark notebooks, Real-Time Intelligence , the Fabric Data Warehouse , event streams and Copilot.
That single pool is what makes Fabric different from buying separate Azure services, because you don’t reserve compute for each tool. Instead, the SKU size sets the ceiling, and your workloads share it.
The Three Things You Pay For A Fabric bill has three main parts. First, the capacity is the compute, charged per capacity-unit hour. OneLake storage then has its own charge per gigabyte per month. Finally, per-user Power BI licenses sit on top for the people who build and share reports.
A handful of optional meters can add to that, such as capacity overage, Autoscale Billing for Spark, SQL database storage and cache storage. The official Microsoft Fabric pricing page lists each of them, and later sections of this guide explain when they matter.
What a Capacity Unit Is in Plain Terms A capacity unit is Microsoft’s measure of compute power across every Fabric engine. For example, an F2 gives you 2 CUs and an F64 gives you 64 CUs. So the number after the F always equals the CU count. When a query, refresh or notebook runs, Fabric records how many CU-seconds it used.
You never pay per query. Instead, you pay for the capacity to be running, whether it is busy or idle. That is why sizing and scheduling matter more than query-by-query tuning when you plan a Fabric budget.
Bursting, Smoothing and Throttling Fabric lets a job run faster than your SKU’s normal limit, which Microsoft calls bursting. It then spreads that usage over future time slots, which is smoothing. Fabric smooths interactive work such as report queries over at least five minutes. It smooths background work such as refreshes and pipelines over 24 hours.
Smoothing is why a small SKU can still finish a large overnight job. It also means a heavy day can leave the capacity paying back yesterday’s usage. When that debt grows too large, throttling starts, following the stages in Microsoft’s throttling documentation .
Up to 10 minutes of future capacity used means no throttling at all. Between 10 and 60 minutes, interactive jobs wait 20 seconds before they start. Between 60 minutes and 24 hours, Fabric rejects interactive jobs such as report visuals. Beyond 24 hours, Fabric also rejects background jobs such as refreshes. Throttling doesn’t raise the bill, but it hurts users. So the practical choice is to size up, spread jobs out or pay for overage, and each of those has a cost you can plan. The Microsoft Fabric capacity planning guide covers the monitoring side in more depth.
Microsoft Fabric Capacity Pricing: The F-SKU Price Table Microsoft prices Fabric capacity per CU-hour, so every SKU costs the same rate times its CU count. In East US and West US 2 the pay-as-you-go rate is $0.18 per CU-hour. A one-year reservation works out to $938 per CU per year. The table below applies those list rates from the Azure Retail Prices API to every SKU, using 730 hours per month.
Table 1: Microsoft Fabric F-SKU capacity pricing, East US list prices
SKU Capacity units Pay-as-you-go per hour Pay-as-you-go per month 1-year reservation per month F2 2 $0.36 $262.80 $156.33 F4 4 $0.72 $525.60 $312.67 F8 8 $1.44 $1,051.20 $625.33 F16 16 $2.88 $2,102.40 $1,250.67 F32 32 $5.76 $4,204.80 $2,501.33 F64 64 $11.52 $8,409.60 $5,002.67 F128 128 $23.04 $16,819.20 $10,005.33 F256 256 $46.08 $33,638.40 $20,010.67 F512 512 $92.16 $67,276.80 $40,021.33 F1024 1,024 $184.32 $134,553.60 $80,042.67 F2048 2,048 $368.64 $269,107.20 $160,085.33 F4096 4,096 $737.28 $538,214.40 $320,170.67 F8192 8,192 $1,474.56 $1,076,428.80 $640,341.33
Source: Azure Retail Prices API, Microsoft Fabric meters for East US, checked September 24, 2026 ($0.18 per CU-hour pay-as-you-go, $938 per CU per year reserved). Monthly figures use 730 hours. Prices vary by region, currency and agreement.
Reading the Price Table Two things stand out. First, the range now runs to F8192, well past the F2048 ceiling that older pricing guides still quote. Second, the price scales in a straight line, so an F128 costs exactly twice an F64 with no volume discount at the top end.
Microsoft also notes that listed prices are estimates, because your agreement, currency and purchase date change the final number. Treat the table as a list-price baseline for budgeting, then confirm with the Azure pricing calculator or your Microsoft account team.
Kanerika Service
Microsoft Fabric Services From Kanerika
Kanerika sizes Fabric capacity from real workload data, reviews Power BI licensing and sets up capacity guardrails so the first invoice matches the budget.
Explore Fabric Services Choosing a Starting SKU F2 to F8 suits proof-of-concept work, a single department’s lakehouse or development and test capacity. Meanwhile, F16 to F32 covers most departmental analytics with scheduled pipelines and a few dozen report creators. Finally, F64 and above is where company-wide reporting, heavier Spark work and large report audiences usually land.
Still, SKU size alone doesn’t decide the bill. Two companies with the same headcount can need very different capacities. Refresh frequency, data volume and query design drive CU usage far more than user counts do.
Pay-As-You-Go vs Reserved Fabric Capacity Microsoft bills pay-as-you-go capacity for the time it runs, with a one-minute minimum, and you can pause it or change its size at any time. By contrast, a reservation commits you to a set number of CUs for one or three years in exchange for a lower rate. Microsoft states that a Fabric capacity reservation can save about 41% compared with pay-as-you-go.
The 60% Break-Even Rule For example, a one-year reservation for an F64 costs about $5,002.67 a month. The same F64 on pay-as-you-go costs $11.52 an hour, so $5,002.67 buys about 434 hours, or roughly 60% of a 730-hour month. If the capacity will run more than about 60% of the time, reserve it.
If it runs only during business hours, pay-as-you-go with a pause schedule is usually cheaper. When a capacity runs 10 hours a day on 22 working days, it runs 220 hours. That is only about 30% of the month and far below the break-even point.
One-Year or Three-Year Reservation The Azure Retail Prices API currently lists the three-year Fabric reservation at $2,814 per CU. That is the same $938 per CU per year as the one-year term. A three-year term buys price certainty for longer, but at today’s list rates it doesn’t buy a deeper discount.
A reservation also doesn’t save money when paused, because you pay for the reserved CUs either way. Many teams reserve the steady baseline, such as the production capacity, and keep development and burst capacity on pay-as-you-go.
What a Reservation Doesn’t Cover Reservation discounts apply to the base capacity only. So Autoscale Billing for Spark and capacity overage both use pay-as-you-go based rates, even on a reserved capacity. Storage is also billed separately.
Microsoft Fabric Pricing by Region Fabric prices vary by Azure region, and you fix the region when you create the capacity. US regions such as East US and West US 2 are among the lowest-cost options. Western Europe costs about 22% more per CU-hour at list price.
Table 2: Microsoft Fabric capacity price by Azure region, USD list prices
Region Pay-as-you-go per CU-hour F64 pay-as-you-go per month 1-year reservation per CU per year F64 reserved per month East US $0.18 $8,409.60 $938.00 $5,002.67 West US 2 $0.18 $8,409.60 $938.00 $5,002.67 Central India $0.20 $9,344.00 $1,042.00 $5,557.33 UK South $0.21 $9,811.20 $1,094.00 $5,834.67 Australia East $0.21 $9,811.20 $1,094.00 $5,834.67 West Europe $0.22 $10,278.40 $1,146.00 $6,112.00
Source: Azure Retail Prices API, checked September 24, 2026. Local-currency prices on the Azure pricing page follow Microsoft’s monthly exchange-rate rules.
So keep the capacity in the same region as your data and users where you can. Moving data between regions can add network costs later, and data residency rules often decide the region before price does.
Microsoft Fabric Licensing: Free, Pro and Premium Per User Capacity pays for compute, but people still need licenses. Fabric has three per-user license types, and the right mix depends on who builds content, who shares it and who only reads it. Microsoft’s Fabric licenses documentation is the reference for everything in this section.
The Free Fabric License Fabric assigns the Microsoft Fabric (Free) license automatically when a user first signs in, as long as Fabric is enabled for the tenant. It lets a user create and share non-Power BI items such as lakehouses , warehouses and notebooks. Those items must live in a workspace backed by an F capacity or a trial capacity.
However, the Free license doesn’t let a user create or share Power BI content outside their own My workspace. It also doesn’t let them view Power BI reports on a capacity smaller than F64.
Power BI Pro and Premium Per User Power BI Pro costs $14.00 per user per month and Premium Per User (PPU) costs $24.00. Both use annual billing, according to Microsoft’s Power BI pricing page . Users who already have Pro, including through Microsoft 365 E5, can step up to PPU with a $14.00 add-on.
Publishing Power BI content to any Fabric capacity also requires a Pro license. PPU adds Premium features for Power BI, such as larger semantic models and 48 refreshes a day. PPU doesn’t provision a Fabric capacity, so a PPU-only organization still can’t create lakehouses, warehouses or notebooks.
Table 3: Which Fabric and Power BI license each user needs
What the user does Capacity below F64 (F2 to F32) F64 or larger Price per user per month Builds lakehouses, warehouses, notebooks or pipelines Free license Free license $0 Creates and shares Power BI reports Power BI Pro or PPU Power BI Pro or PPU $14 (Pro) or $24 (PPU) Only views Power BI reports Pro, PPU or individual trial Free license with Viewer role $0 on F64+, $14 below Needs Premium features without any capacity Premium Per User Premium Per User $24 Tries Fabric before buying 60-day trial capacity (F4 or F64) Not needed $0 for 60 days
Source: Microsoft Learn, Understand Microsoft Fabric licenses and capacity, and the Microsoft Power BI pricing page, checked September 24, 2026.
Teams that also carry Microsoft 365 and Azure agreements can fold this into a broader Microsoft licensing optimization review. The Power BI Premium vs Pro licensing guide goes deeper on the per-user choices for Power BI-only teams.
The F64 Threshold: When Free Report Viewers Pay Off F64 is the most important line in the Fabric price list. In other words, on F64 and larger capacities, users with a Free license and a Viewer role can open Power BI reports. On F32 and smaller, every viewer needs a Pro, PPU or individual trial license.
As a result, that rule turns SKU choice into a licensing decision. Moving from a reserved F32 to a reserved F64 adds about $2,501 a month. At $14 per Pro seat, that equals about 179 viewers, so the math turns on audience size. If more than roughly 179 people only read reports, F64 is often cheaper than licensing each of them.
Where the F64 Math Breaks Down The comparison changes when viewers are few or the workload is small. A team with 40 report readers will almost always spend less on an F16 plus Pro seats. Report creators still need Pro on every SKU, so they stay on the bill either way.
Capacity headroom matters too. Hundreds of new viewers also add real query load. So an F64 that barely fits the data work today may need to grow once you turn on free viewing.
Case Study
60% Less Reporting Effort Across 3 SAP Systems on Fabric
Kanerika unified three SAP systems into one governed Fabric platform across development, test and production, cutting time spent per reporting cycle by 60%.
Read the Case Study → Is Microsoft Fabric Free? Microsoft Fabric is a paid platform, with two free entry points. For example, the Free license lets people work in Fabric at no per-user cost. However, it only works inside a paid F capacity or a trial capacity.
For evaluation, the Microsoft Fabric trial gives 60 days of access at no cost. Microsoft configures the trial capacity as either an F4 or an F64. It comes with a Power BI individual trial for users who don’t already have PPU. Microsoft notes that trial users on 4 CUs may be able to request 64.
A trial is a good place to measure how many CUs your real workloads use. However, it isn’t a place to run production. Non-Power BI items that aren’t moved to a paid capacity before the trial ends are removed under Microsoft’s retention policy.
Costs Beyond Capacity: OneLake Storage, AI and Other Meters Many budget surprises in Fabric come from meters outside the capacity price. Each one is small on its own. Together, however, they can move a monthly bill well past the SKU price when nobody tracks them.
OneLake Storage Tiers OneLake charges for storage per GB per month, separately from capacity. In East US, hot storage lists at $0.026 per GB, cool storage at $0.019 and cold storage at $0.004, based on the retail price feed. Cool and cold tiers were added in July 2026, so older guides that quote a single storage rate are out of date.
OneLake cache, which covers KQL cache and Activator retained data, also costs more at $0.20 per GB. Business continuity and disaster recovery (BCDR) storage costs more than standard storage, at $0.0468 per GB for the hot tier. Power BI licensing already covers the data stored for Power BI import semantic models.
Mirroring, SQL Database and Deleted Workspaces Mirroring gives free replica storage equal to the SKU number in terabytes, so an F64 includes 64 TB of free mirroring storage. That allowance doesn’t come with the trial, and OneLake charges start once you exceed it or pause the capacity.
A SQL database in Fabric stores data at $0.25 per GB per month, plus $0.10 per GB for backups. Deleted workspaces also keep billing for storage during their retention period, which admins can set from 7 to 90 days.
What Copilot, AI and Data Agents Cost Copilot in Fabric doesn’t carry a separate per-user license. Instead, it consumes CUs from a capacity, just like a refresh or a query. The retail price feed lists several AI meters. Copilot and AI usage, ontology work behind Fabric IQ and operations agents all bill at the same $0.18 per CU-hour as every other workload.
Admins can route that usage to a dedicated Fabric Copilot capacity , which can be as small as an F2. That keeps AI experiments from throttling production reports and puts AI spend on its own line. The same capacity also carries Fabric data agents when users on smaller SKUs call them.
Capacity Overage and Autoscale Billing for Spark Fabric capacity overage is a newer option that lets a critical capacity keep running instead of throttling. Microsoft bills excess usage at $0.54 per CU-hour in East US, which is three times the pay-as-you-go rate, and reservation discounts don’t apply. Admins can set a 24-hour spending limit, and Fabric checks that limit every five minutes.
Autoscale Billing for Spark moves Spark jobs to serverless compute billed at the pay-as-you-go CU rate. It helps with unpredictable data engineering work, but you still need a base capacity for everything else. Networking charges are listed as coming soon, and Microsoft has said it will give at least 90 days of notice before they start.
Moving From Power BI Premium P SKUs to Fabric F SKUs Microsoft no longer sells Power BI Premium P SKUs. According to the Premium migration FAQ , each P SKU subscription retires at the end of its current agreement term. Expiring Enterprise Agreements can’t add or renew P SKU capacity.
Once a P SKU ends, content stays available for a 30-day grace period. From day 31, Fabric throttles interactive work, and from day 91 it rejects every operation. Migration isn’t automatic, so each Fabric workspace needs to move to a new F capacity by hand.
A P1 maps to an F64, since both provide 64 capacity units and both allow free viewers. Still, the bigger change is how you pay. P SKUs sat on a Microsoft 365 commitment, while F SKUs are Azure resources you can pause, resize and reserve. Teams also moving Azure Data Factory or Synapse workloads can follow the Azure to Fabric migration guide .
Webinar
Migrate to Microsoft Fabric 5X Faster With FLIP
Kanerika’s office-hours session on moving Azure Data Factory, Synapse and SSIS workloads onto Fabric capacity with its FLIP migration accelerators.
Watch the Session → Worked Microsoft Fabric Cost Examples The three examples below use the East US list prices from Table 1. Pro is priced at $14 per user and hot OneLake storage at $0.026 per GB. Each example shows how the parts of a bill add up, so treat them as planning estimates, not quotes.
Table 4: Worked Microsoft Fabric monthly cost examples at East US list prices
Scenario Capacity Power BI licenses OneLake hot storage Estimated monthly total Small team, 8 report users F4 reserved, $312.67 8 Pro, $112.00 500 GB, $13.00 About $438 Mid-size, 250 users on F32 F32 reserved, $2,501.33 250 Pro, $3,500.00 5 TB, $133.12 About $6,134 Mid-size, 250 users on F64 F64 reserved, $5,002.67 30 Pro creators, $420.00 5 TB, $133.12 About $5,556 Enterprise, 3,000 viewers F128 reserved $10,005.33 + F16 dev/test pay-as-you-go 220 hours $633.60 + F8 Copilot capacity reserved $625.33 120 Pro creators, $1,680.00 40 TB, $1,064.96 About $14,009
Estimates only. They use list prices from Table 1. Power BI Pro is priced at $14 per user and hot storage at $0.026 per GB, with 1 TB counted as 1,024 GB. They exclude taxes, discounts, overage and other optional meters.
Small Team Getting Started A five-person data team runs one lakehouse, a few pipelines and reports for eight Pro users. Together, an F4 reservation, eight Pro seats and 500 GB of storage come to about $438 a month. If the same team only works business hours, an F4 on pay-as-you-go paused outside 10-hour weekdays costs about $158 for capacity instead.
Mid-Size Company With 250 Report Users This company has 30 report creators and 220 people who only read reports. On an F32 reservation, every one of the 250 users needs Pro, and the total reaches about $6,134 a month. On an F64 reservation, only the 30 creators need Pro, and the total drops to about $5,556, while the extra CUs add headroom.
Enterprise Estate An enterprise runs an F128 reservation for production, an F16 on pay-as-you-go for development and testing, and an F8 reservation as a dedicated Copilot capacity. When you add 120 Pro creators, 3,000 Free viewers and 40 TB of hot storage, the list-price total is about $14,009 a month. The 3,000 viewers add nothing to the license line because production runs above F64.
Case Study
40% Lower Costs: SSIS to Microsoft Fabric Migration
Kanerika automated a large enterprise’s SSIS pipeline migration to Microsoft Fabric, cutting infrastructure and maintenance costs by 40% and speeding up data processing by 30%.
Read the Case Study → How to Size a Fabric Capacity Before You Buy Sizing is the step that decides whether a Fabric budget holds. In fact, guessing from headcount is the most common sizing mistake. It leads to an F64 nobody needs, or an F16 that throttles every Monday morning.
List the workloads you will run, such as Power BI models, pipelines, Spark jobs, warehouses and real-time analytics, with their data volumes and refresh schedules. Then run the numbers through Microsoft’s Fabric SKU estimator to get a starting SKU, and treat the output as a first guess. Next, prove it on a trial or a pay-as-you-go capacity for two to four weeks with real data and real users. Read the Microsoft Fabric Capacity Metrics app to see CU use over time, throttling events and the items that use the most. The Fabric capacity planning guide walks through that monitoring in detail. Split production from development and testing, and promote content through deployment pipelines , so experiments can’t slow down the reports people depend on. Reserve only the baseline that runs most of the month, and keep the rest flexible. Two companies with the same number of users can still land two SKUs apart. One refreshes a large model every 15 minutes while the other refreshes once a night. That difference shows up in CU-hours long before it shows up in user counts.
How to Lower Your Microsoft Fabric Bill Most Fabric savings come from matching capacity to the hours and the load you actually have. The graphic and the list below cover the moves that work for most estates.
Pause non-production capacity. A development capacity that runs 10 hours a weekday costs about 30% of a 24-hour one. Microsoft’s pause and resume guide notes that pausing bills any smoothed usage still outstanding.Reserve the steady baseline only. Then reserve production capacity that runs above the 60% break-even, and handle peaks with a temporary pay-as-you-go scale-up.Fix the heaviest items first. The Capacity Metrics app ranks items by CU use, and one inefficient refresh or notebook often uses more than dozens of reports.Stagger background jobs. Spreading refreshes and pipelines across the night also keeps smoothing from piling usage into the business day.Use surge protection. Surge protection lets admins cap background jobs sooner and set a CU limit per workspace, which keeps one team’s spike from blocking others.Tier cold data. Because cold data rarely changes, it can move to the cool or cold OneLake tier. That cuts its storage price by up to 85% compared with hot storage.Charge back by workspace. Showing each business unit its share of CU usage makes teams tidy up unused items and schedules on their own. It also fits naturally into a wider Fabric governance model.Broader Azure savings habits apply too, and the Azure cost optimization playbook covers them. Organizations still planning a move can check whether they qualify for Microsoft partner funding. Kanerika also covers that in how to reduce Fabric migration costs with Microsoft funding .
How Fabric Pricing Compares With Synapse, Databricks and Snowflake Fabric prices compute as always-on capacity, which suits teams that want one predictable monthly number. Azure Synapse Analytics bills each service separately, such as dedicated SQL pools and Spark pools, so costs are spread across several meters.
Databricks bills in Databricks Units on top of the cloud compute it runs on, and Snowflake bills in credits for warehouses that suspend when idle. So both models reward bursty workloads that switch off between jobs. Fabric rewards steady shared use across many workloads, especially Power BI. A reserved F64 at about $5,003 a month covers every Fabric workload around the clock. By contrast, a warehouse that suspends when idle only costs money during the hours it runs.
The cheapest platform depends on workload shape more than on list price. The Databricks vs Snowflake vs Fabric comparison walks through that decision. Meanwhile, the Fabric vs Power BI guide also helps Power BI-only teams decide whether they need Fabric at all.
How Kanerika Keeps Fabric Spend Predictable Kanerika is a Microsoft Solutions Partner for Data and AI with the Analytics Specialization and a Microsoft Fabric Featured Partner. Its teams size, migrate and run Fabric estates for enterprises moving from Power BI Premium, Azure Data Factory, Synapse, SSIS and other platforms.
How the Work Runs Assess. Kanerika inventories current Power BI licenses, P SKU terms, refresh schedules and Azure data services, so the Fabric plan starts from real usage.Size. The team runs representative workloads on trial or pay-as-you-go capacity and reads the Capacity Metrics app before recommending a SKU and a reservation split.Migrate. FLIP , Kanerika’s migration accelerator platform, automates much of the move from Azure data services, SSIS and Informatica into Fabric, and Kanerika’s Azure to Fabric migration services handle the rest.Govern. Separate production and development capacities, surge protection, workspace chargeback and a Copilot capacity keep costs visible after go-live.Enable. Report creators and admins learn to read the Metrics app, so the capacity stays right-sized as usage grows.Results From Kanerika Fabric Projects For a large enterprise with complex SSIS pipelines, Kanerika automated the move to Microsoft Fabric using PySpark notebooks and Power Query. The SSIS to Fabric case study reports a 40% reduction in infrastructure and maintenance costs, 30% faster data processing and 25% less manual maintenance.
A global thermal-management manufacturer ran its order journey across three disconnected SAP systems. Kanerika then built one governed Fabric platform across development, test and production, connecting about 35 source tables and 50+ KPIs. The SAP order-lifecycle case study reports a 60% cut in time spent per reporting cycle.
Microsoft’s own FoodPharma customer story describes how FoodPharma worked with Kanerika to unify six systems on Fabric. Cross-functional reporting went from two business days to 90 minutes, and the BI team got back roughly 15 hours a week.
What Kanerika Teams Watch For The most common cost mistake Kanerika sees is a capacity sized for data engineering with no plan for report viewers. The second is a reservation bought before anyone has measured a month of real usage. Both are cheap to avoid at the assessment stage, although they are expensive to unwind after a one-year commitment.
A third pattern is AI usage landing on the production capacity. Picture Copilot and data agents sharing an F SKU with finance reports. Then a burst of AI questions at month end can push that capacity into throttling. A small dedicated Copilot capacity keeps both the risk and the cost on a separate line that finance can see.
Kanerika also builds tooling that runs inside Fabric. In addition, Karl, its AI data insights agent, and the Azure to Fabric Migration Accelerator are both native Microsoft Fabric workloads. So teams can use them from the same capacity they already pay for. Learn more on the Kanerika Microsoft Fabric services page .
Talk to Kanerika
Get a Fabric Capacity and Licensing Review
Talk to Kanerika’s Fabric team about SKU sizing, reservation timing and Power BI license mix before you buy or renew.
Book a Meeting → Wrapping Up Microsoft Fabric pricing comes down to three numbers you can plan. The capacity follows a simple $0.18 per CU-hour list rate in most US regions, with about 41% off for a reservation. Licenses then depend on the F64 line, which decides whether report viewers pay. Storage and optional meters such as Copilot, overage and Spark autoscale fill in the rest. Measure real usage on a trial or pay-as-you-go capacity first, reserve only the steady baseline, and check the F64 math against your viewer count. That order keeps the first Fabric invoice close to the budget.
Frequently Asked Questions
How much is Microsoft Fabric per month? Microsoft Fabric starts at about $262.80 a month for an F2 capacity on pay-as-you-go in East US. An F64 costs about $8,409.60 a month at the same rate. A one-year reservation lowers those figures by about 41%. Storage and any Power BI Pro licenses are billed on top of the capacity price.
How much is a Microsoft Fabric subscription? Fabric is priced in three parts. You pay for a capacity by the hour or through a reservation, then add OneLake storage and Power BI licenses for report creators. The smallest capacity, an F2, lists at about $263 a month on pay-as-you-go in East US. Larger SKUs scale in direct proportion.
How much does a Microsoft Fabric license cost? The Free Fabric license costs nothing and is assigned when a user first signs in. Power BI Pro costs $14 per user per month and Premium Per User costs $24, both paid yearly. The capacity itself is billed separately in Azure, starting at $0.18 per capacity unit per hour in East US.
How does Microsoft Fabric pricing compare to Databricks? Fabric bills a fixed capacity that every workload shares, so the monthly cost stays predictable. Databricks bills Databricks Units on top of the cloud virtual machines it runs on, and clusters can shut down between jobs. Steady, Power BI-heavy estates often suit Fabric. Bursty engineering workloads can cost less on Databricks.
What is the equivalent of Microsoft Fabric? The closest equivalents are Databricks, Snowflake and Google BigQuery, each paired with a BI tool. None bundles storage, engineering, warehousing, real-time analytics and BI under one capacity the way Fabric does. That bundling is why Fabric pricing is one capacity bill, while the others spread costs across several services and licenses.
Is Microsoft Fabric a replacement for Synapse? Microsoft positions Fabric as the successor to Azure Synapse Analytics for most new analytics work. Synapse bills each pool and service separately. Fabric bills one capacity for data engineering, warehousing and Power BI together. Teams moving from Synapse should compare their current pool spend against the F SKU that fits their measured workload.
What's the point of Microsoft Fabric? Fabric puts data integration, engineering, warehousing, real-time analytics and Power BI on one platform with one storage layer called OneLake. For budgeting, that means one capacity to buy instead of several Azure services. It also means Power BI and data workloads share compute, so planning capacity carefully matters more than on separate tools.
Why use Microsoft Fabric? Teams use Fabric to cut the number of tools, copies of data and bills they manage. A single capacity covers pipelines, lakehouses, warehouses and reports, and OneLake keeps one copy of the data. Organizations with large Power BI audiences also gain free report viewing once they run an F64 or larger capacity.
What problems does Microsoft Fabric solve? Fabric addresses scattered analytics tools, duplicate data copies and separate security settings across services. It stores data once in OneLake and lets every engine read it. On the cost side, it replaces several service bills with one capacity. That makes spending easier to track, as long as the capacity is sized from real usage.
What are the limitations of Microsoft Fabric? Fabric runs only in Azure regions, so multi-cloud estates still need other tools. Capacity is billed while it runs, even when idle, unless you pause it. Shared capacity means one heavy job can throttle reports for everyone. Teams with very spiky workloads may find per-job billing on other platforms cheaper.
Is Microsoft Fabric the future? Microsoft is retiring Power BI Premium P SKUs and directing customers to Fabric F SKUs. It keeps adding Fabric workloads such as SQL database, Fabric IQ and data agents that bill against the same capacity. For Power BI-centered organizations, planning a Fabric budget is now part of normal licensing work, not a side project.
Is Microsoft Fabric free or paid? Microsoft Fabric is paid. Production work needs an F capacity billed in Azure, and Power BI creators need Pro licenses. Microsoft offers a Free per-user license and a 60-day trial capacity. The Free license only works inside a paid or trial capacity, so it lowers user costs without removing the capacity cost.
What are some Microsoft Fabric alternatives? Common alternatives are Databricks, Snowflake, Google BigQuery and Amazon Redshift, usually paired with Power BI or another BI tool. Each uses a different pricing unit, such as Databricks Units, Snowflake credits or per-query scanning. Compare them on your own workload shape, since steady and bursty workloads favor different pricing models.
Can I get Microsoft Fabric for free? You can use Fabric free for 60 days through the trial capacity, which runs as an F4 or F64. The Free user license is also permanent and costs nothing. After the trial, keeping your items requires a paid capacity. Items that aren’t moved to one are removed under Microsoft’s retention policy.
How do I purchase Microsoft Fabric? You buy Fabric capacity in the Azure portal. Create a Fabric capacity resource in your Azure subscription, then choose an F SKU and a region. You can buy a reservation for that capacity separately in Azure. Power BI Pro and PPU licenses are bought through the Microsoft 365 admin center.
Is Microsoft Fabric included in Office 365? No. Fabric capacity isn’t included in Microsoft 365 or Office 365 plans and is billed in Azure. Microsoft 365 E5 includes Power BI Pro, which covers the per-user license for building and sharing reports. You still need an F capacity to run Fabric items such as lakehouses, warehouses and notebooks, and that capacity is billed separately.
What license is required for Microsoft Fabric? Every Fabric user gets a Free license on first sign-in, which covers lakehouses, warehouses and notebooks on an F or trial capacity. Anyone who creates or shares Power BI content needs Pro or PPU. Report viewers need Pro below F64 and only the Free license with a Viewer role on F64 or above.
Is Microsoft Fabric pay as you go? Yes. Fabric capacity can run on pay-as-you-go, billed for the time it runs with a one-minute minimum. You can pause it, resume it or change its size whenever you need, which suits development work and seasonal peaks. Capacity that runs more than about 60% of the month usually costs less on a one-year reservation.
Is Microsoft Fabric expensive? Fabric can be cheaper or more expensive than separate tools depending on sizing. An F2 costs about $263 a month on pay-as-you-go, while an F64 costs about $8,410. Costs rise fastest when capacity is oversized, left running unpaused or paired with Pro licenses for hundreds of report viewers below F64.
How to choose Microsoft Fabric capacity? Start with Microsoft’s Fabric SKU estimator using your data volumes, refresh schedules and user counts. Then test on a trial or pay-as-you-go capacity for two to four weeks. Read the Capacity Metrics app for CU use and throttling. Pick the smallest SKU that stays healthy, and check the F64 viewer rule.
How much does an F64 capacity cost? An F64 costs $11.52 an hour on pay-as-you-go in East US, which is about $8,409.60 for a 730-hour month. A one-year reservation brings it to about $5,002.67 a month. Prices are higher in some regions. For example, the same F64 costs about $10,278 a month on pay-as-you-go in West Europe, before any discounts.
Do report viewers need a Power BI Pro license on Fabric? It depends on the capacity size. On F2 through F32, every person who views Power BI content needs Pro, PPU or an individual trial. On F64 and larger, viewers with only a Free license and a Viewer role on the workspace can open reports. Creators need Pro on every size.
Can I pause a Fabric capacity to save money? Yes, a pay-as-you-go F capacity can be paused in the Azure portal, and billing for compute stops while it is paused. Pausing adds any smoothed usage still outstanding to your bill. Content on a paused capacity is unavailable to users. For that reason, pause only development, test or out-of-hours capacity.
What happens when a Fabric capacity runs out of CUs? Fabric first borrows up to 10 minutes of future capacity with no slowdown. Past that point, new interactive jobs wait 20 seconds before they start. After an hour of borrowed capacity, report queries are rejected, and after 24 hours refreshes are rejected too. Resizing, pausing or spreading jobs out clears the backlog.
How much does Copilot in Microsoft Fabric cost? Copilot in Fabric is billed through capacity units, with no separate per-user Copilot fee. Its usage is metered at $0.18 per CU-hour on pay-as-you-go in East US, the same rate as other Fabric workloads. Admins can send Copilot usage to a dedicated Fabric Copilot capacity, which can be as small as an F2.
How much does OneLake storage cost? OneLake storage is billed per GB per month outside the capacity price. In East US, hot storage costs $0.026, cool storage $0.019 and cold storage $0.004 per GB. Cache storage costs $0.20 per GB. Mirrored replicas get free storage up to the SKU number in terabytes, such as 64 TB on an F64.
What happens to Power BI Premium P SKUs? Microsoft no longer sells P SKUs, and each one retires at the end of its current agreement term. After it ends, content stays available for 30 days, is throttled from day 31 and is rejected from day 91. Customers need to buy an F SKU and reassign their workspaces manually.
How long is the Microsoft Fabric free trial? The Fabric trial lasts 60 days unless you cancel it sooner. The trial capacity runs as an F4 or an F64 and includes a Power BI individual trial for users without PPU. Before it ends, move workspaces to a paid capacity, or non-Power BI items are removed under the retention policy.
Is a three-year Fabric reservation cheaper than a one-year reservation? The two terms currently cost the same per year. The Azure Retail Prices API lists a three-year Fabric reservation at $2,814 per capacity unit, which is $938 a year. The one-year term is also $938. Both save about 41% compared with pay-as-you-go, so the longer term mainly buys price certainty.
Why does Microsoft Fabric pricing differ by region? Microsoft sets a separate Fabric capacity rate for each Azure region. In US dollar list prices, pay-as-you-go costs $0.18 per CU-hour in East US, $0.21 in UK South and $0.22 in West Europe. The region is fixed when you create a capacity. Data residency rules often decide it before price does.
