Power BI and Microsoft Fabric pricing in 2026: what you actually pay

Power BI Pro is $14 per user per month and Premium Per User is $24, but neither number decides your bill. The F64 capacity threshold does, because that is where report viewers stop needing a licence. Here is the cost model, with the figures Microsoft publishes and the ones it does not.

How much does Power BI cost in 2026?

Power BI Pro is $14 per user per month on an annual commitment. Power BI Premium Per User is $24 per user per month on the same terms. Microsoft Fabric capacity is billed separately through Azure, per second with a one-minute minimum, and Microsoft does not print a flat monthly figure for it on the pricing page.

That is the honest answer to the list-price question. It is also the least useful part of the decision, because for any organization past about thirty report readers the bill is decided by something else: whether you buy per user or buy capacity, and where the line between those two sits for your headcount.

What this is

A cost model for Power BI and Microsoft Fabric, written from procurement conversations rather than from a vendor page. It covers the licence tiers, the capacity ladder, the one licensing rule that moves the maths more than any discount, and the costs that land after the licence.

What this is not

A price list we invented. Where Microsoft publishes a number, it is quoted here with the terms attached. Where Microsoft does not publish a number, this page says so instead of estimating one. Fabric capacity pricing is regional and Microsoft directs buyers to the Azure pricing calculator or a sales quote, so any article showing you a confident F64 monthly figure is guessing on your behalf.

The per-user tiers, and what each one actually buys

Free. Assigned automatically the first time a user signs into the Fabric portal. It lets someone create non-Power BI Fabric items in a workspace backed by an F or Trial capacity. It does not let them create Power BI content outside their own My workspace, and on a capacity below F64 it does not let them view Power BI content either.

Pro, $14 per user per month. The licence that lets a person publish and share Power BI content. Every organization using Power BI needs at least one. On F capacities smaller than F64, every person who so much as opens a report needs one too.

Premium Per User, $24 per user per month. Adds most of the Premium feature set on a per-user basis: larger semantic models, more frequent refreshes, the XMLA endpoint. Microsoft's own guidance is that PPU is the cost-effective route when you need Premium features for fewer than 250 users.

The trap in PPU is what it does not include. PPU does not provision a Fabric capacity. If your plan involves lakehouses, warehouses, notebooks or pipelines, PPU will not run them, and you need an F capacity regardless of how many PPU seats you have bought. We have watched a team buy PPU across a department expecting it to cover a Fabric build, and it did not.

The capacity ladder, and why there is no sticker price

Fabric capacity comes as F SKUs bought through Azure, from F2 up to F8192, measured in capacity units. F64 is the rung that maps to the old P1, at 8 v-cores.

Azure bills these per second with no commitment, and you can pause a capacity when it is not in use, which is a genuine lever if your workload is business-hours only. A yearly reservation is available and Microsoft states it saves 40.5% against pay-as-you-go, so the reservation decision is worth real money once your usage pattern is steady.

Pricing is regional, and Microsoft does not publish the per-SKU figures on the Fabric pricing page. Get your number from the Azure pricing calculator signed in under your own agreement, or from a quote. Do not budget from a blog post, including this one.

One planning note that costs people money: Microsoft is consolidating purchase options and retiring the Power BI Premium per-capacity P SKUs, and its own documentation now tells new and existing customers to consider F capacities instead. If you are sitting on a P SKU renewal, that is the conversation to have now rather than at renewal minus two weeks.

The F64 cliff

This is the number that decides the shape of your bill, and it is not a price.

On a Fabric capacity of F64 or larger, a user with only a Free licence and a viewer role on the workspace can view Power BI content. Below F64, every one of those viewers needs Pro, PPU or an individual trial.

So the cost curve is not smooth. Under F64 you are paying for capacity and paying $14 a head for everyone who reads a report. At F64 and above the per-viewer charge disappears and you are paying for capacity alone.

That is why a 400-person reporting audience and a 40-person one are different purchases rather than different sizes of the same purchase. It is also why the honest breakeven question is not "is capacity cheaper than per-user", it is "how many people read reports, and does that number justify climbing to F64".

Working the breakeven

Run it with your own numbers rather than a rule of thumb.

Per-user route, annual:

Pro seats × $14 × 12 + PPU seats × $24 × 12 (only for the people who need Premium features) + any F capacity you still need for non-Power BI Fabric workloads

Capacity route, annual:

F SKU cost from the Azure calculator, reserved or pay-as-you-go + Pro or PPU seats for the people who CREATE and share content, which the capacity does not cover + $0 for viewers, but only at F64 and above

Two mistakes to avoid in that calculation. First, count active readers, not licensed users; the gap between the two is usually 30% or more and it is the single biggest source of over-buying we see. Second, remember that capacity never removes the creator licences. Authors still need Pro or PPU no matter how large the capacity is, and budgets that assumed otherwise come up short.

What lands on the bill after the licence

• Capacity sizing gets bought twice. Most teams overshoot at purchase and then discover throttling under real concurrency anyway, because sizing follows refresh windows and query patterns rather than user count. Plan to re-size after a month of real telemetry.
• Power BI cost overruns are usually semantic model overruns. A model that will not fit a smaller SKU is a modelling problem before it is a licensing problem, and it is often cheaper to fix the model than to buy the next rung.
• Frequent refreshes of large models are what push a capacity up a tier. Incremental refresh is the cheapest performance work available and it is routinely skipped.
• On-premises data sources need a gateway, which needs a server, patching and someone who owns it.
• Capacity monitoring, workspace governance and access review are a real part of someone's week once you are past a couple of hundred users.

Where the estimates go wrong

The pattern we see most often is a business case built on licensed seats at list price, compared against a competitor quote built the same way, with neither side counting the model rebuild that the migration actually requires. Both numbers are wrong and the decision gets made on them anyway.

The second pattern is a Fabric purchase justified by the F64 viewer rule when the organization has 60 report readers. At that size the per-user route is usually cheaper, and the capacity gets bought for a headroom that never arrives.

The buy-side checklist

1. Count report readers active in the last 30 days, not licences assigned. 2. Separate creators from viewers. Creators need Pro or PPU either way. 3. Decide whether you need non-Power BI Fabric workloads at all. If not, PPU may cover you without any capacity. 4. Price the F SKU in the Azure calculator under your own agreement, both reserved and pay-as-you-go. 5. Check whether your largest semantic model fits the SKU you are pricing before you commit. 6. If you hold P SKUs, get the migration path costed now, given the retirement direction. 7. Ask what the number looks like in year two, after the model grows and the audience does.

What we do

We run this calculation as part of Power BI and Fabric engagements, and we do it before anyone signs anything. Bring your active reader count, your refresh schedules and the size of your largest semantic model, and we will tell you which side of the F64 line you belong on and what the model work costs. Sometimes the answer is that your current licensing is already right and the problem is the model, in which case that is what we will say.

If you are running the same comparison against Tableau, the Tableau pricing guide covers that side with the same structure, and Tableau vs Power BI covers the platform decision rather than the invoice. Our Power BI consulting and Microsoft Fabric pages cover the delivery work.

Frequently asked questions

How much does Power BI Pro cost in 2026?

Power BI Pro is $14 per user per month, paid yearly, per Microsoft's published pricing. It lets a user publish Power BI content and share it, and on Fabric capacities below F64 every person viewing that content needs a Pro licence as well.

What is the difference between Power BI Pro and Premium Per User?

Premium Per User is $24 per user per month, paid yearly, and adds most Power BI Premium features on a per-user basis: larger semantic models, more frequent refreshes and XMLA endpoint access. Microsoft's guidance is that PPU is more cost effective than a capacity when fewer than 250 users need Premium features. PPU does not provision a Fabric capacity, so it will not run lakehouses, warehouses or notebooks.

How much does Microsoft Fabric capacity cost?

Microsoft does not publish per-SKU dollar figures on the Fabric pricing page. Capacity is bought through Azure as F SKUs from F2 to F8192, billed per second with a one-minute minimum, with regional pricing and an optional yearly reservation that Microsoft states saves 40.5% against pay-as-you-go. Price your specific SKU in the Azure pricing calculator signed in under your own agreement.

Do report viewers need a Power BI Pro licence?

It depends entirely on capacity size. On a Fabric capacity of F64 or larger, users with a Free licence and a viewer role on the workspace can view Power BI content. On any F SKU smaller than F64, each viewer needs Pro, PPU or an individual trial. That threshold, not the licence price, is what decides the shape of most Power BI bills.

When does a Fabric capacity beat per-user licensing?

The breakeven is driven by how many people read reports. Below F64 you pay for capacity and still pay per viewer, so capacity adds cost without removing any. At F64 and above viewers are covered by the capacity, so the more readers you have, the better capacity looks. Creators still need Pro or PPU in both models.

Is Power BI Premium being retired?

Microsoft is consolidating purchase options and retiring the Power BI Premium per-capacity P SKUs, and its documentation directs new and existing customers to consider Fabric F capacity subscriptions instead. Existing P SKU customers on an active Enterprise Agreement can still run them, but the migration is worth costing before renewal rather than at it.

Does Premium Per User include Microsoft Fabric?

No. PPU provides Power BI Premium features on a per-user basis but does not provision a Fabric capacity. Creating or running non-Power BI Fabric items such as lakehouses, warehouses and notebooks requires an F capacity or a Trial capacity, regardless of how many PPU licences you hold.

How does Power BI pricing compare to Tableau?

The per-user entry point is much lower: Power BI Pro at $14 per user per month against Tableau Cloud Creator at a considerably higher per-seat figure. The comparison stops being simple at scale, because Power BI shifts to capacity pricing while Tableau stays per-seat, and because the migration cost of moving semantic models between them usually outweighs the first-year licence difference either way.