Power BI Pricing and Licensing 2026: Every SKU, Real Costs

Power BI costs are set by two independent decisions: how many people need a per-user license, and whether you buy capacity. Pro is $14.00 and Premium Per User is $24.00 per user per month on annual billing, as published at time of writing (July 2026). Capacity changes the math entirely — at Fabric F64 and above, viewers need no paid license. EPC Group is a Houston-based Microsoft consulting firm operating since 1997, with six Microsoft Solutions Partner designations and 216+ M&A tenant migrations covering 1.83 million users.

Quick facts

The 2026 SKU table

Capacity units (CUs) are the unit of compute. The v-core column exists only to translate legacy Power BI Premium sizing; it is a compute comparison, not a statement of licensing equivalence.

Microsoft publishes F SKU dollar prices on the Azure pricing page and calculator only, and the rate varies by region. This article quotes no F SKU dollar figure, because no single figure is correct everywhere. The arithmetic below takes your quoted regional price as its input.

The Five-Layer Power BI Cost Stack

Most licensing quotes are wrong because they price one layer and forget four. EPC Group uses a five-layer stack on every Power BI commercial model. Price all five, or the number you take to finance will move.

1. Viewer layer. People who only consume. $0 on F64+, or $14.00 each below it. This layer decides whether you buy capacity at all.
2. Author layer. People who publish, share, or manage workspaces. Always Pro or PPU, on every SKU. Capacity does not make authors free.
3. Capacity layer. The F SKU, the pay-as-you-go/reservation decision, overage exposure, and the dev/test capacity nobody budgets for.
4. Connectivity layer. Gateways: the VMs, the cluster members for high availability, and sizing headroom. The 30-refresh concurrency limit means a second cluster, not a bigger box.
5. Governance layer. Purview, tenant settings enforcement, certification workflow, and the standing effort to keep it true — the line item most often discovered after signature.

Layers 4 and 5 are absent from every page currently ranking on “power bi pricing.” They are why quoted budgets slip.

The break-even: Pro versus F64, with the arithmetic shown

The only economic question that matters at scale: at what viewer count does capacity cost less than per-viewer licences?

Let C be your quoted monthly F64 cost in your Azure region. Authors pay $14.00 either way, so they cancel out of both sides. The comparison is viewers-avoided against capacity cost.

Break-even viewers = C ÷ $14.00

Below that viewer count, keep buying Pro. Above it, capacity is cheaper — assuming F64 has the compute headroom for the workload, which is a separate test covered further down.

Substitute your own regional quote. The formula does not change.

Worked model at three organisation sizes

Assume a common enterprise shape: 10% authors, 90% viewers. All figures use the verified $14.00 Pro price.

250-user organisation — 25 authors, 225 viewers

• All-Pro: 250 × $14.00 = $3,500/month = $42,000/year
• On F64: (25 × $14.00) + C = $350 + C per month
• Capacity must cost under 225 × $14.00 = $3,150/month to win
• Verdict: do not buy capacity. No realistic F64 quote clears $3,150. This organisation buys Pro, and buys PPU only for the specific users who need Premium features. Microsoft’s own scenario guidance points the same way: PPU for populations under roughly 250 Power BI users.

2,000-user organisation — 200 authors, 1,800 viewers

• All-Pro: 2,000 × $14.00 = $28,000/month = $336,000/year
• On F64: (200 × $14.00) + C = $2,800 + C per month
• Capacity must cost under 1,800 × $14.00 = $25,200/month to win
• Verdict: capacity wins decisively on licensing. Every plausible F64 quote clears the bar. The question is no longer whether to buy capacity but which size — and whether one F64 carries 2,000 users’ refresh and interactive load. Baseline before you buy.

15,000-user organisation — 1,500 authors, 13,500 viewers

• All-Pro: 15,000 × $14.00 = $210,000/month = $2,520,000/year
• On capacity: (1,500 × $14.00) + C = $21,000 + C per month
• Licensing avoided: 13,500 × $14.00 = $189,000/month = $2,268,000/year
• Verdict: capacity is not optional, and F64 is almost certainly the wrong size. The constraint flips from licensing to compute: F64 caps a single semantic model at 25 GB and delivers 1,536 CU hours per day. Size on measured consumption, not headcount. The gap between F128 and F256 is a rounding error against $2.27M/year of avoided Pro licences — buy the size that performs, not the cheapest.

The pattern: the break-even sits in the mid-hundreds of viewers. Below ~400 viewers, buy per-user licences. Above ~1,000, buy capacity and spend the remaining effort on sizing. Between the two, run the formula with your actual quote.

Which SKU do you actually need — decision tree

The hidden costs

Six line items that do not appear on the Microsoft pricing page and do appear on your invoice.

1. Gateways. Gateways run on machines you pay for, and high availability means a cluster. The gateway enforces a concurrency limit of 30 refreshes; past it you get “too many refreshes occurring concurrently,” and the remedy is another cluster member. Microsoft also recommends separate gateways for Import and DirectQuery — a second estate, not a second service. See EPC Group’s Power BI gateway guidance.

2. Capacity overage. Capacity overage (preview) prevents throttling by paying off excess usage — and Azure bills it at three times pay-as-you-go rates. Keep the rolling 24-hour limit below one-third of daily CU hours; beyond that you are paying more than a scale-up costs. Overage limits also consume Fabric quota at 1/24th of the limit set.

3. Dev/test capacity. Production capacity should not carry development load. Standard patterns: one production F SKU plus a smaller capacity paused outside business hours, or two right-sized F SKUs. Never split a capacity already near 100% utilisation — that manufactures overloads on both halves.

4. Purview and governance tooling. Sensitivity labels, DLP, catalogue, and audit evidence are licensed and operated separately. Budget the operating effort as well as the licence — see best data governance consulting firms.

5. Storage and networking. A reservation covers capacity usage only. Storage and networking are charged at pay-as-you-go rates regardless, and the discount applies hourly with no carry-forward — unused reserved hours are lost, not banked.

6. Report Server. Entitlement after P SKU retirement comes from a Fabric capacity reservation or SQL Server Enterprise Edition with Software Assurance. Pay-as-you-go F capacity does not carry it. Verify before cancelling the P SKU; the key then sits on the Fabric capacity tab under Capacity settings.

Pay-as-you-go versus reservation

Reservations are bought in one-CU increments for one or three years, scoped to an Azure region. Microsoft states the saving as approximately 41% against pay-as-you-go.

Four rules decide the answer:

• Reserve only measured baseline. The discount applies hourly and does not carry over; a reservation larger than your running capacity wastes the excess every unmatched hour.
• Reserve after stabilising. Track spend daily for 30 days with Microsoft Cost Management, then reserve the floor.
• Reservations remove pause/resume value. If capacity genuinely idles overnight, pausing pay-as-you-go may beat reserving 24/7. Model both.
• Reservations do not auto-renew unless set to. On expiry, workloads continue and quietly revert to pay-as-you-go rates.

What breaks — failure modes

What changed in 2026

• P SKU retirement moved from advisory to operational. Microsoft no longer sells new P SKUs, and customers with expiring Enterprise Agreements or Microsoft Cloud Agreements cannot add or renew P capacity. Migration is manual — Fabric does not decommission the P SKU when you provision an F SKU.
• The retirement now has teeth with dates attached: 30-day grace, throttled interactive access days 31–90, all operations rejected from day 91 until you migrate or delete the capacity.
• Capacity overage (preview) replaced Autoscale as the surge mechanism on F SKUs, with an explicitly published 3× billing multiplier and an admin-set rolling 24-hour limit.
• Embedded consolidated into F. Power BI Embedded is included with every F SKU, removing the separate EM/A purchase for organisations moving to Fabric. Existing EM and A SKUs remain valid.
• Sovereign clouds are carved out. Fabric is not yet available there, so P SKUs remain supported in those environments pending separate guidance.
• The April 2025 per-user price change is now fully through the renewal cycle, so 2026 renewals are the first at steady state. Separately, monthly billing on annual-term subscriptions carries a 5% premium over annual upfront billing.

Frequently asked questions

Where EPC Group fits

EPC Group has run Microsoft data platform commercial modelling since 1997 — six Microsoft Solutions Partner designations, 11,000+ engagements, consecutive G2 Leader quarters. Errin O’Connor, Founder and Chief AI Architect, is a 4× bestselling author.

Related reading: Power BI consulting services · Power BI cost and licensing guide · Power BI Premium vs Premium Per User · Power BI Premium · Microsoft Fabric consulting services · Fabric consulting guide · Microsoft Fabric vs Databricks · Snowflake to Fabric migration · Power BI vs MicroStrategy · Microsoft 365 E3 vs E5 · Microsoft 365 Copilot pricing and licensing · Top Power Platform consulting firms · Power BI consulting in Houston · Power BI consulting in Phoenix · Microsoft Frontier Company.

Get the model run against your numbers

Bring your user population, Azure region, and renewal date. EPC Group runs the Five-Layer Cost Stack and the break-even formula against your actual quote, and returns a sized recommendation with the arithmetic shown. Start at Power BI consulting services.

Sources and verification

1. Microsoft — Power BI pricing (Pro and PPU list prices, annual billing)
2. Microsoft Learn — Understand Microsoft Fabric licenses (SKU/CU table, F64 free-viewer rule)
3. Microsoft Learn — Power BI Premium to Microsoft Fabric migration overview
4. Microsoft Learn — P SKU to F SKU migration decision guide (retirement timeline, P→F mapping)
5. Microsoft Learn — Premium to Fabric migration FAQ (PPU not retired, EM/A out of scope)
6. Microsoft Learn — Migrate workspaces from Power BI Premium to Microsoft Fabric
7. Microsoft Learn — Capacity overage (preview) in Microsoft Fabric (3× billing, CU hours/day)
8. Microsoft Learn — Save costs with Microsoft Fabric Capacity reservations
9. Microsoft Learn — Plan your capacity size (CU per SKU, 30-second evaluation period)
10. Microsoft Learn — The Fabric throttling policy
11. Microsoft Learn — Capacity and SKUs in Power BI embedded analytics (A SKU behaviour, 25 GB model limit)
12. Microsoft Learn — Direct Lake overview (per-SKU guardrails and max memory)
13. Microsoft Learn — Partner Center March 2025 announcements (1 April 2025 price change, 5% monthly premium)
14. Microsoft Learn — Power BI service features by license type
15. Microsoft Learn — Troubleshoot the on-premises data gateway (30 concurrent refresh limit)
16. Microsoft Azure — Microsoft Fabric pricing (pay-as-you-go and reservation, regional variance)
17. Microsoft Learn — Microsoft Fabric capacity planning guide