Power BI Free vs Pro vs Fabric sounds like a simple pricing question. In practice, it is one of the most misunderstood decisions in the Microsoft data stack. Organisations overpay for licences nobody uses. Others buy a Fabric capacity that is one size too small and still have to license every report viewer. Meanwhile, Microsoft has retired new Power BI Premium capacity purchases, so many older licensing guides are now simply wrong.

This guide is for BI leaders, IT managers and finance teams who sign off on analytics spend. First, we explain what each tier really allows. Next, we compare them side by side with verified limits and prices. Finally, we share a practical way to choose, plus the mistakes that cost organisations the most. Every fact here comes from Microsoft's own documentation and pricing page, checked in November 2026.

Power BI Free vs Pro vs Fabric: The Short Answer

Power BI licensing works on two separate axes. The first is the per user licence: Free, Pro or Premium Per User (PPU). It decides what each person can do. The second is capacity: a Fabric F SKU that your organisation buys once and shares. It decides how much compute you have, which Fabric workloads you can run and, above a certain size, whether viewers need a paid licence at all.

That last point matters. Buying Fabric does not remove the need for Pro. Report creators still need a Pro or PPU licence to publish Power BI content, whatever capacity you own.

What Power BI Free Really Gives You

Power BI Desktop is a free download. Anyone can use it to connect to data, build a model and design a report, saved as a .pbix file. So for learning, prototyping and personal analysis, Free is genuinely useful.

The limits appear the moment you want to share. With a free licence, you can publish only to My workspace, which is for personal use. You cannot create or contribute to shared workspaces, share content with colleagues or build apps. However, there is one important exception. If your organisation hosts content on an F64 or larger Fabric capacity, a free user with the Viewer role can open and interact with those reports. In other words, Free becomes a consumption licence once the right capacity is in place. Microsoft lists the details in Power BI features available to free users.

That consumption licence is more capable than many people assume. With the right admin settings, free viewers on F64 or larger can export reports to PDF and PowerPoint, print pages and subscribe themselves to email updates. With Build permission, they can also use Analyze in Excel. They cannot, however, edit, reshare or create content in shared workspaces.

Two practical rules also catch people out. First, signing up for Power BI needs a work or school email address; personal accounts such as Gmail cannot be used directly. Second, administrators can switch off self-service sign-up, purchases and trials, so if the "Start trial" button is missing, check tenant settings before anything else.

Power BI Pro and Premium Per User

Power BI Pro is the standard paid licence. It lets people publish to shared workspaces, collaborate and share reports with other licensed users. Microsoft lists it at $14 per user per month, paid yearly, and it is already included in Microsoft 365 E5 and Office 365 E5. Pro semantic models are limited to 1 GB, with up to eight scheduled refreshes per day.

Premium Per User (PPU) adds most Premium features on a per person basis. Microsoft lists it at $24 per user per month, paid yearly. Alternatively, users who already have Pro or an E5 licence can step up with a $14 add-on. In return, PPU raises the model size limit to 100 GB and allows 48 scheduled refreshes per day. It also adds deployment pipelines and XMLA endpoint read and write access. Microsoft positions PPU as cost effective for fewer than 250 users.

Storage and refresh time limits differ too. Pro allows 10 GB of storage per user, while PPU and capacity allow up to 100 TB. In addition, a scheduled refresh on shared capacity must finish within two hours, whereas PPU and capacity allow five. For teams with slow source systems, that time-out is often the real reason to upgrade.

Power BI Free vs Pro vs Fabric: Side by Side

The table below brings the limits together. Prices are Microsoft's US list prices for annual billing and vary by country, currency and agreement.

Fabric SKUs Explained: Sizes, Prices and Limits

Fabric capacity is sold in F SKUs. The number after the F is the count of capacity units (CUs), so an F64 has 64 CUs and twice the compute of an F32. Sizes start at F2 and go up to F8192. You pay through Azure, either pay as you go by the hour or with a one or three year reservation.

The table below uses Microsoft's published East US prices in US dollars, taken from the Azure Retail Prices API in November 2026. Pay as you go is $0.18 per CU per hour, and a one year reservation is $938 per CU per year. Monthly figures assume 730 hours. Prices differ by region and currency, so treat these as a guide, not a quote.

Three details in that table drive most decisions. First, F64 is the smallest SKU where free licence users can view Power BI content, which is why it appears in so many proposals. Second, a reservation cuts the price by 40.5%, but you commit for the full term. Third, OneLake storage is billed separately per GB, on top of capacity, and it keeps billing even while a capacity is paused. For reference, Microsoft maps a legacy P1 to F64 by capacity units. For the full limits per SKU, see Microsoft's capacity and SKU documentation.

Power BI Free vs Pro vs Fabric: Which Is Better, and Why?

There is no single winner. Instead, each option is the better choice for a different shape of organisation. Pro wins on simplicity for small and mid-sized teams. PPU wins when a small group needs bigger models. Fabric wins once you need a data platform or a large read-only audience. The real question is what Fabric gives you that per user licences cannot.

What Fabric gives you that Pro and PPU do not

• A full data platform: lakehouses, warehouses, notebooks, pipelines and real-time analytics, all reading one copy of data in OneLake.
• Direct Lake: Power BI reads Delta tables in OneLake directly, so large models do not need a full import refresh.
• Dedicated compute: your workloads run on capacity you own, rather than shared Microsoft compute.
• Free licence viewers from F64: read-only users no longer need a paid licence each.
• Embedding included: every F SKU covers Power BI Embedded, so you do not need a separate A or EM SKU. When you embed reports in your own app for external customers, those viewers need no licence at all.
• On-premises reporting: a Fabric capacity reservation includes rights to Power BI Report Server. Pay as you go capacity does not.
• Copilot: Copilot runs on paid F2 or larger capacity, while Pro and PPU workspaces cannot run it on their own.

Where Fabric saves money

• Viewer licences: above roughly 358 read-only users, a reserved F64 costs less than buying Pro for each of them.
• Reservations: committing for a year saves 40.5% against pay as you go.
• Pausing: a pay as you go capacity stops compute billing while paused, which suits development and test environments.
• Fewer separate products: embedding, Report Server rights and data engineering tools come under one capacity bill.
• Less duplicated data: with Direct Lake, models read data where it already lives instead of keeping a refreshed copy.

Where Fabric costs more

• Small teams: 150 users on Pro cost $2,100 a month, while a reserved F64 alone costs about $5,003.
• Always-on capacity: capacity bills whether or not anyone uses it. Moreover, pausing it makes its content unavailable, so production capacity usually runs around the clock.
• Creator licences remain: everyone who publishes Power BI content still needs Pro or PPU, on top of the capacity.
• Storage and growth: OneLake storage is billed separately, and an undersized capacity gets throttled, which pushes you to a bigger SKU.

"Fabric is not a cheaper Power BI. It becomes the cheaper option once you have enough readers, or enough data work, to fill a capacity."

How Many Users Before Fabric Pays Off?

The maths is simpler than most proposals make it look. Creators need Pro or PPU either way, so they cancel out. Therefore, the real comparison is the capacity price against Pro for every read-only user. A reserved F64 costs about $5,003 a month, which equals Pro for about 358 viewers. On pay as you go, the break-even rises to about 601 viewers.

Two cautions apply. First, if your users already hold Microsoft 365 E5, Pro is included, so capacity saves nothing on viewer licences. Second, F64 must actually handle your workload. A thousand heavy users on large models may need F128, so size it with a trial and the Capacity Metrics app first. Similarly, PPU for everyone matches a reserved F64 at around 208 users, which lines up with Microsoft's guidance that PPU suits fewer than 250 users.

Want Copilot without moving everything to Fabric?

A Fabric Copilot capacity is a useful middle path. An administrator designates one paid F2 or larger capacity and assigns users to it. From then on, their Copilot usage in Pro and PPU workspaces, and in Power BI Desktop, bills to that single capacity. As a result, you can roll out Copilot to a group without reassigning every workspace.

Try before you buy

The Fabric trial lasts 60 days and runs as an F4 or F64 equivalent. It also comes with a licence similar to PPU, so you can test PPU features at the same time. However, it does not include Copilot or data agents. When the trial ends, workspaces revert to Pro and items such as notebooks and pipelines become inactive. Their content stays in OneLake for seven days, so assign those workspaces to paid capacity before then. Finally, note that Microsoft no longer sells new Power BI Premium P SKUs, so F SKUs are now the capacity option for new purchases.

Power BI Free vs Pro vs Fabric: How to Choose

Most organisations end up with a mix. Typically, a small group of creators holds Pro or PPU, while capacity serves everyone else. The decision path below helps you find your starting point.

In practice, the answer usually follows the shape of the team. The table below shows where each common situation tends to land.

Six Licensing Mistakes We See Most Often

Licensing errors rarely look dramatic. Instead, they show up quietly as wasted spend or blocked users. These are the six that come up most.

• Treating PPU as Premium capacity: teams buy PPU expecting free users to view reports. They cannot, because PPU content requires PPU for every viewer.
• Buying an F SKU below F64 for viewers: smaller capacities run Fabric workloads well. However, every Power BI viewer still needs Pro or PPU.
• Forgetting creator licences: capacity covers compute, not authoring. Anyone who publishes Power BI content still needs Pro or PPU.
• Judging Copilot on a trial: Fabric trials do not include Copilot. As a result, teams wrongly conclude it is unavailable to them.
• Paying twice for Pro: Microsoft 365 E5 and Office 365 E5 already include Pro, yet organisations sometimes buy it again.
• Letting a P SKU lapse: when a Premium capacity agreement ends, there are 30 days of grace, throttled access from day 31 and blocked operations from day 91. So buy and test the F SKU before the old agreement ends.

Ultimately, the cheapest licence is the one that matches how people actually use your reports. A clear view of creators, viewers and workloads usually settles the Power BI Free vs Pro vs Fabric question within a week. For the full rules, see Microsoft Fabric licenses on Microsoft Learn.

Numlytics specialises in Power BI consulting, Power BI governance and Microsoft Fabric migration. Speak with a certified consultant to find the right licensing mix for your organisation.