For most small companies, Power BI Pro is the right license. At $14 per user per month (Microsoft's list price, paid yearly), it covers building, sharing and refreshing reports for a team, with models up to 1 GB and 8 scheduled refreshes a day. Move to Premium Per User ($24 per user per month) when your models outgrow 1 GB or you need refreshes more often than every few hours, and everyone who uses those reports can be on PPU. Look at a Fabric capacity when you want many people to view reports without each needing a paid license, or when you're building a data warehouse. Free viewers need an F64 capacity or larger, which is well beyond most small-business budgets.
What each option is
Power BI Pro is a per-user license. Everyone who builds, shares or views shared content needs one. It's included in Microsoft 365 E5 and Office 365 E5, so check whether you already have it.
Premium Per User (PPU) is also per user. It includes everything in Pro, plus bigger models, more refreshes and most Premium features. The catch: content in a PPU workspace can only be viewed by other PPU users, unless the workspace sits on a Premium or Fabric capacity.
Fabric capacity (F SKUs, from F2 upward) is a pool of computing power your whole organization shares, billed through Azure by the hour or with a one- or three-year reservation. It runs Power BI along with the rest of Microsoft Fabric: data warehouses, lakehouses, pipelines and notebooks. People still need a Pro license to build and publish. Below F64, viewers still need Pro too.
Prices here are Microsoft's US list prices on the Power BI pricing page as of September 2026. Your price can differ by region, agreement or reseller.
The differences that matter to a small company
| Pro | Premium Per User | Fabric capacity | |
|---|---|---|---|
| Cost | $14 per user per month | $24 per user per month | Hourly or reserved, per capacity size, plus Pro for authors |
| Who can view shared reports | Other Pro users | Other PPU users | Pro users; free users too at F64 and above |
| Max model size | 1 GB | Up to 100 GB | Depends on capacity size |
| Scheduled refreshes per day | 8 | 48 | 48 |
| Max refresh time | 2 hours | 5 hours | 5 hours |
| Data warehouse, lakehouse, pipelines | No | No | Yes |
A few other points that come up:
- Incremental refresh works on all three. Only the real-time DirectQuery part of it needs PPU or a capacity.
- Larger models need the large semantic model storage format, which you switch on in the model's settings on PPU or a capacity.
- Fabric capacity can be paused in the Azure portal to stop the ongoing charges, but content on a paused capacity isn't available to anyone until you resume it. That works for a dev or test capacity, not for reports people open every morning.
When Pro is enough
Pro fits if these are all true:
- Your models are under 1 GB, or can be after you trim them. (Most small-business models can. Our post on memory limit errors shows how.)
- Refreshing every few hours during the working day is fresh enough.
- The people who look at reports can each have a license, or there aren't very many of them.
- Your data already lives in a database or a system with a connector, so you don't need a warehouse.
For a 10-person company where everyone uses Power BI, that's $140 a month.
Signs you've outgrown Pro
- A model won't fit in 1 GB even after cleanup, and it's not something incremental refresh fixes.
- 8 refreshes a day isn't enough. Someone needs numbers that are an hour old, not four.
- Refresh takes close to 2 hours.
- You want features Pro doesn't include, such as connecting other tools to your model with read and write access through the XMLA endpoint.
If only a few people build and use those heavier reports, PPU is usually the cheapest step up. Remember that everyone viewing that content needs PPU too.
When a Fabric capacity makes sense
- Lots of viewers, few authors. If 150 people need to look at reports and 5 build them, F64 or above lets the viewers use free licenses. Do the math against Pro for everyone. For most small companies Pro still wins.
- You need a place to store and shape data, not just report on it. For example, pulling data from several systems into one warehouse, or moving off an aging Azure SQL database. That's what Fabric is for.
- You want predictable capacity for many models and refreshes rather than paying per user.
Get the capacity size wrong and reports slow down or get throttled at busy times, so size it from real usage rather than guessing. Microsoft's Fabric capacity estimator is a reasonable starting point, and you can resize later.
What we'd do
- Start on Pro.
- Trim your models and set up incremental refresh before you pay for more.
- Move a small group to PPU if their reports really need bigger models or more refreshes.
- Look at Fabric once you need a warehouse, or once free viewers would save more than the capacity costs.
If you're deciding between Import and DirectQuery at the same time, read DirectQuery vs Import first, because it changes how much capacity you need.
When to call someone
Licensing mistakes are expensive in both directions: paying for capacity you don't use, or rebuilding reports because Pro couldn't hold them. It's worth a second opinion before you buy a Fabric capacity, before moving a team to PPU, or when a report has hit a limit and you're not sure whether it's the license or the model.
If one report is broken right now because of a limit, our $299 Power BI fix covers it, and you don't pay if we can't fix it. For a licensing and setup review, see Power BI advisory.
Sources: Microsoft, Power BI pricing; Microsoft Learn, Power BI licenses, Premium Per User FAQ, Large semantic models, Data refresh in Power BI, Pause and resume your Fabric capacity.