Power BI Embedded licensing in 2026 is a capacity-purchase decision priced through Azure A-SKUs (A1 at ~$735/month through A8 at ~$94,800/month) intended for ISV "embed for your customers" scenarios where end users are outside your tenant. The A-SKU surface is materially overlapping with the Microsoft Fabric F-SKUs that have absorbed Power BI Premium per capacity. For internal-app embedding inside your tenant the right SKU is Power BI Pro or PPU plus a Premium / Fabric capacity, not Power BI Embedded. Most enterprises buying A-SKUs are buying the wrong product for an internal use case, and most ISVs buying F-SKUs are missing the auto-pause economics that A-SKUs offer. The 2026 task is to map embed scenario against the correct SKU and stop paying for the wrong capacity.
What Power BI Embedded actually licenses
Power BI Embedded is a capacity-billing product priced through Azure that licenses Power BI reports, dashboards, and dataset queries when those reports are surfaced inside a customer-facing application built by an ISV or a software vendor. The "embed for your customers" scenario is the licensing case Power BI Embedded was designed for — an external software vendor builds a SaaS application and surfaces Power BI content to end users who do not have Power BI Pro or Premium per User licences and who typically do not exist in the ISV’s Microsoft tenant. The A-SKU capacity carries the licensing entitlement; the end users do not require Power BI licences of their own.
That commercial design matters because the second use case — "embed for your organisation" — is the one most enterprises mistakenly buy A-SKUs for. When you embed Power BI content inside your own internal application surfaced to your own employees, you are not licensing Power BI through the A-SKU capacity. You are licensing it through each employee’s Power BI Pro or Premium per User entitlement plus a Premium / Fabric capacity for the back-end data refresh. The A-SKU was never the right purchase for the internal scenario, and the procurement waste compounds when an internal application is sized against an A4 or A5 capacity instead of the Fabric F-SKU plus per-user mix the deployment actually needs.
A-SKU pricing tiers in 2026
The A-SKU lineup mirrors the original Power BI Premium P-SKU lineup, with the same v-core counts and roughly matching monthly prices. The advantage A-SKUs carry over F-SKUs is Azure billing — pause-and-resume billing is the differentiator. A-SKUs can be stopped and started through the Azure portal or REST API, and billing pauses while the capacity is stopped. For ISVs running embed-for-customers scenarios with predictable business hours and known low-traffic windows, A-SKU pause-and-resume can drop effective monthly cost by 30–55% against an equivalent F-SKU running 24/7.
| A-SKU | v-cores | Memory | Approx. monthly | Best fit |
|---|---|---|---|---|
| A1 | 1 | 3 GB | $735 | Dev / test, low-volume ISV demo tier |
| A2 | 2 | 5 GB | $1,470 | Small ISV deployment, <500 concurrent users |
| A3 | 4 | 10 GB | $2,940 | Mid ISV, 500–2,000 concurrent users |
| A4 | 8 | 25 GB | $5,895 | Equivalent to P1 Premium capacity |
| A5 | 16 | 50 GB | $11,790 | Equivalent to P2 |
| A6 | 32 | 100 GB | $23,580 | Equivalent to P3 |
| A7 | 64 | 200 GB | $47,160 | Equivalent to P4 |
| A8 | 128 | 400 GB | $94,320 | Equivalent to P5 |
The exact monthly prices vary by Azure region and by the regional uplift Microsoft applies through the EA. Customers buying A-SKUs through an EA with a Microsoft Azure Consumption Commitment (MACC) can apply A-SKU spend against the MACC, which is the structural reason A-SKUs end up cheaper than F-SKUs for many EA customers — the F-SKU equivalent does not count toward the MACC in the same way.
A-SKU vs Fabric F-SKU for embedded analytics
The Fabric F-SKU surface that absorbed Power BI Premium per capacity in the 2026 migration overlaps materially with the A-SKU surface. The differences that drive the procurement decision:
- Billing model. A-SKUs are pause-and-resume Azure billing. F-SKUs are committed monthly capacity with optional reservation. For 24/7 ISV deployments the F-SKU 1-year reservation is typically 40–42% cheaper than pay-as-you-go A-SKU running continuously. For business-hours-only ISV deployments the A-SKU pause-and-resume model is materially cheaper.
- MACC eligibility. A-SKU consumption counts toward Microsoft Azure Consumption Commitment burn. F-SKU consumption traditionally has not, though Microsoft has signalled a path for Fabric to count toward MACC by mid-2026 — verify with the LSP before assuming.
- Fabric workloads. F-SKUs unlock the full Fabric suite (Data Engineering, Data Science, Synapse Data Warehouse, Real-Time Analytics, Data Activator) on the same capacity. A-SKUs are Power BI-only. If the analytics platform is being built across Fabric workloads, F-SKU is the only commercial path.
- Sovereign clouds and BYOK. A-SKU availability in sovereign clouds (Government, Germany, China) is broader than F-SKU availability at this writing. ISVs with sovereign-cloud customer requirements still need A-SKUs in regions where F-SKUs have not landed.
The decision matrix: if the deployment is pure Power BI embed-for-customers with predictable low-traffic windows, A-SKU plus pause-and-resume scripting is the cheapest path. If the deployment is multi-workload Fabric or runs 24/7 with steady traffic, F-SKU with 1-year reservation is cheaper. The Fabric P-to-F migration playbook covers the broader F-SKU economics; for embedded specifically the A-SKU branch remains live.
Embed for your organisation: the wrong-SKU trap
The single most expensive Power BI Embedded mistake is buying an A-SKU for an internal app. When an enterprise builds an internal application — an HR analytics dashboard, a finance KPI portal, a supply-chain tracker — and surfaces Power BI content inside that app to its own employees, the licensing requirement is not the A-SKU capacity. Each employee viewing the content needs a Power BI Pro licence ($14/user/month) or a Premium per User licence ($24/user/month), and the back-end refresh and dataset hosting needs a Premium / Fabric capacity. The A-SKU does not grant per-user entitlement for internal users; Microsoft will accept the configuration during deployment and then surface the licensing shortfall during a compliance review or audit.
The audit defence is straightforward: enumerate every embedded Power BI scenario in the estate, classify each as embed-for-customers (external) or embed-for-organisation (internal), and verify the SKU mix matches. The most common procurement waste we find is an A4 or A5 capacity ($5,895–$11,790/month) running an internal HR dashboard with 800 employees who already hold Power BI Pro entitlements through their M365 E5 — the A-SKU is unnecessary and the deployment should run on the existing Fabric capacity the tenant already owns.
If your A-SKU consumption sits inside your own tenant and the viewers all hold corporate email addresses, the A-SKU is almost certainly the wrong product. Re-platform onto Fabric F-SKU plus per-user Pro/PPU and capture the saving in the next EA amendment.
EA negotiation levers for Power BI Embedded
- MACC application. A-SKU spend applied to MACC reduces the effective unit cost of the capacity through the EA commitment burn. Negotiate the MACC sizing to include the embedded scenario.
- Reserved capacity discount. A-SKUs are eligible for Azure reservation discounts at 1-year and 3-year tenors when run continuously. Reservation discounts run 32–38% at 1-year and 52–59% at 3-year on the eligible v-cores. If the embed scenario runs 24/7, reservation is the lowest-effort cost lever.
- F-SKU migration credit. For ISVs migrating from A-SKU to F-SKU at Microsoft’s prompting, negotiate transition credits to bridge the period during which Fabric features the ISV does not need are bundled into the F-SKU price. The leverage is real because the migration is Microsoft-led.
- Autoscale and burst. A-SKUs support autoscale-to-A-tier-up behaviour. Negotiate the autoscale ceiling explicitly in the deployment runbook so a misconfigured spike does not double the monthly bill silently.
- Pause-and-resume runbook. The largest A-SKU cost lever is pause-and-resume against business hours. Build the runbook before the renewal and surface the projected savings in the cost case to the LSP.
Anonymised case study: $360K Power BI Embedded rebuild
A 4,500-employee SaaS vendor surfacing analytics to its customer base ran two A5 capacities ($23,580/month combined, $283K annualised) plus an internal HR dashboard on a third A4 ($5,895/month, $70,740 annualised). Total embedded spend: $354K annualised. We audited the deployment: the customer-facing A5 capacities ran 24/7 against 11am–7pm UTC peak traffic with negligible weekend volume. The internal HR dashboard was deployed on an A4 against 800 internal viewers who already held M365 E5 with Power BI Pro entitlement and against a tenant Fabric F32 capacity that was 51% utilised. We rebuilt: the customer-facing A5s moved to A4 with pause-and-resume scripting against business-hours-only windows (effective monthly: $3,275 each, $78,600 combined annualised); the internal HR dashboard moved to the existing Fabric F32 capacity at zero incremental capacity cost. Total annualised cost after rebuild: $78,600. Net annualised saving: $275K. The MACC sizing was reduced by $250K against the same EA term and redirected toward Azure infrastructure spend at the renewal.
Power BI Embedded is one of the cleanest cost-lever surfaces in the Power Platform portfolio — the SKU economics are well documented, the pause-and-resume mechanism is standard, and the misconfiguration patterns are repeatable. Pair the embedded rebuild with the wider M365 mix rebuild and a clear position on the 2026 EA tier-collapse landscape, and Power BI Embedded stops being a Premium-shaped commitment and starts being a managed Azure consumption line. For the broader Power Platform context see our Power Platform Licensing Guide.