Power BI Licensing Explained: Pro vs Premium Per User vs Fabric Capacity (2026)
What each licence costs, what it unlocks, who needs one, and the viewer count where capacity becomes cheaper than per-user licences.
Core Engineering Takeaways
- •Power BI Pro costs $14 per user per month and Premium Per User (PPU) $24, both billed annually. Fabric capacity is billed per capacity, not per user.
- •Pro limits a semantic model to 1 GB and 8 scheduled refreshes a day. PPU raises that to 100 GB and 48 refreshes.
- •PPU content can only be viewed by other PPU users. That one rule decides most Pro-vs-PPU questions.
- •Free-licence viewers are only allowed on F64 or larger. An F64 costs about $5,003/month reserved, so it beats Pro at roughly 358 viewers.
1. The four ways to license Power BI
Power BI licensing looks complicated because it mixes two models: per-user licences (Free, Pro, Premium Per User) and capacity (Fabric F SKUs, plus legacy Premium P SKUs and Embedded A SKUs). Every user has a per-user licence; a capacity is something a workspace runs on. What someone can do depends on both.
- Free: build content in My workspace and view content that sits on an F64+ (or P) capacity. A free user cannot share or view shared content on anything smaller.
- Pro — $14 per user per month: publish to shared workspaces, share, and view content other Pro users publish. Models up to 1 GB, 8 scheduled refreshes a day. Pro is included in Microsoft 365 E5.
- Premium Per User — $24 per user per month: everything in Pro plus Premium features — models up to 100 GB, 48 refreshes a day, the XMLA endpoint and deployment pipelines — but only for people who also have PPU.
- Fabric capacity (F SKUs): you rent compute for a workspace, from F2 ($263/month pay-as-you-go) upwards. Authors still need Pro or PPU to publish and share. From F64, viewers can use a free licence.
Microsoft no longer sells new Power BI Premium P SKUs; existing ones move to Fabric F SKUs when their term ends. Power BI Embedded A SKUs are a separate Azure product for embedding reports in your own application, covered in the Embedded pricing guide.
2. Pro vs Premium Per User: the sharing rule decides
On features alone, PPU looks like an easy upgrade: 100× the model size, six times the refreshes, XMLA and pipelines for about 70% more per user. The catch is in Microsoft's own wording — with PPU, content can only be shared with other PPU users, unless the workspace sits on a Premium or Fabric capacity.
So the real question is not which licence is better but who looks at the reports:
- A small team that builds and consumes its own reports (finance, a data team): PPU is often the best value. Everyone is an author anyway.
- A few authors and many readers: PPU becomes expensive fast, because every reader also needs PPU. Pro for everyone, or a capacity, usually wins.
- A model above 1 GB, or more than 8 refreshes a day: Pro cannot do it. First check whether the model actually needs to be that large — removing unused high-cardinality columns often brings it back under the limit (see 7 common semantic model problems). If it does, choose PPU or capacity.
Two PPU details people miss: a Pro user added to a PPU workspace still cannot open its reports, and moving a workspace from PPU back to Pro fails if any model is over the Pro limits.
3. When capacity becomes cheaper than per-user licences
Capacity pays off when you have many viewers, because from F64 upwards viewers no longer need a paid licence. Authors still do.
At US East list prices, an F64 costs $8,410/month pay-as-you-go or $5,003/month with a one-year reservation. Divide by the Pro price and the break-even is roughly 358 viewers on a reservation, or 601 on pay-as-you-go.
Worked example: 10 authors, 150 viewers
- Everyone on Pro: 160 × $14 = $2,240/month.
- Everyone on PPU (needed if the content lives in a PPU workspace): 160 × $24 = $3,840/month.
- F64 reserved + Pro for authors: $5,003 + 10 × $14 = $5,143/month.
At 150 viewers, Pro for everyone is clearly cheapest. The picture flips somewhere past 358 viewers — or earlier if you need capacity features anyway (larger models, Fabric items such as lakehouses, Direct Lake, or Copilot).
Smaller F SKUs (F2–F32) are useful for Fabric workloads and Copilot, but they do not remove the viewer licence requirement. A common and costly mistake is buying an F32 expecting free viewers.
4. A quick decision guide
- Under ~50 people, models under 1 GB, refresh a few times a day? Pro for everyone.
- A small, self-contained analyst team that needs big models, XMLA or pipelines? PPU for that team.
- Hundreds of read-only viewers? F64+ capacity with Pro for authors — and price a reservation once usage is stable.
- You need Fabric items (lakehouse, warehouse, notebooks) or Copilot but have few viewers? A small F SKU for the Fabric workloads, Pro for users.
- Showing reports to customers inside your own app? That is embedding — see the Embedded guide; per-user licences are not the right model.
Mixed setups are normal: most mid-sized organisations end up with Pro for everyone, plus one capacity for the workspaces with the widest audience.
5. Costs that do not show up on the licence page
- Capacity sizing mistakes. An undersized capacity throttles and an oversized one wastes money every month. Size from real usage — see P1 to F64 right-sizing.
- OneLake storage is billed separately from Fabric compute.
- Copilot consumes capacity units from the same capacity your reports use, so heavy Copilot use can affect report performance.
- Licence drift. Pro licences assigned to people who left, or PPU bought for a pilot and never removed. A quarterly licence review usually pays for itself.
Prices in this guide are US East list prices checked on 2026-10-10; they vary by region, currency and agreement. Confirm current figures on Microsoft's Power BI pricing page and the licence feature comparison.
Not sure which licence mix fits?
Compare F SKU cost against Pro licences for your own numbers, or get a written recommendation with the migration report.
Get the next Engineering Note
Practical DAX, semantic model and Fabric notes. 1–2 a month, unsubscribe any time.