Microsoft Power Platform Licensing: Deep Dive on Power Apps, Power Automate, and Hidden Costs

The Seeded Entitlement Trap

Every user with an M365 E1, E3, E5, or E7 licence receives a seeded Power Platform entitlement. This entitlement allows users to run Power Apps canvas apps that connect to Microsoft 365 data sources — SharePoint lists, Teams channels, Excel files stored in OneDrive, and other M365-native connectors — without any additional Power Apps licence. For many organisations, this seeded capability is sufficient for basic internal productivity applications.

The trap is triggered the moment an application connects to a premium connector. Premium connectors include Dataverse (Microsoft's low-code database), Dynamics 365 data sources, non-Microsoft SaaS APIs (Salesforce, ServiceNow, SAP), SQL Server, and a growing list of enterprise system connectors. As soon as any user of a seeded-entitlement app accesses content from a premium connector, every user who runs that application requires a Power Apps Premium licence at $20 per user per month — or a Power Apps Per App licence, which was retired from the price list for new customers in January 2026.

This entitlement boundary is the single most common cause of unexpected Power Platform spend in enterprise environments. Development teams build applications on the seeded entitlement, test against SharePoint or Teams data, and then connect to a downstream production system — Dynamics 365, an SQL database, or a third-party API — without flagging the licence impact. The first True-Up after that premium connector goes live triggers a licence retroactive charge for every user who accessed the application during the EA period.

Power Apps Licensing in 2026

Power Apps Premium: $20 Per User Per Month

Power Apps Premium is the standard per-user licence for any user who needs to run or create applications that use premium connectors or Dataverse. At $20 per user per month, it provides unlimited app access — a user can run as many Power Apps as required with a single Premium licence. This makes Premium straightforwardly cost-effective for users who need access to multiple applications across different departments.

The economics change when an application only needs to reach a small subset of users. For 50 users accessing a single app that uses a premium connector, the monthly cost is $1,000. For 500 users accessing the same app, the monthly cost is $10,000. This is where the recently retired Per App plan had appeal — at $5 per user per app per month, it was more cost-effective for users accessing only one or two premium apps.

Per App Plan Retirement (January 2026)

Microsoft retired the Power Apps Per App plan for new customers on 2 January 2026. This was the $5 per user per app per month option that allowed targeted licensing for single-application deployments. Existing EA customers who had Per App licences on their EA schedule prior to the retirement can continue using and renewing them through their current agreement term. For any new Power Platform deployment in 2026, Per App is no longer available — the choice is between the seeded M365 entitlement, Power Apps Premium at $20 per user per month, or Power Apps for specific scenarios such as Dynamics 365 qualifying users.

The Per App retirement is commercially significant. It removes the low-cost entry point for targeted deployments and forces organisations that previously used Per App to migrate to Premium at a 4x price increase per user per app, or to restructure application governance to stay within seeded entitlement boundaries by avoiding premium connectors.

Power Automate Licensing

Power Automate has its own parallel licensing structure that mirrors Power Apps in its seeded-versus-premium split. Users with M365 licences can create and run basic automated flows between M365 services — sending Teams messages based on SharePoint events, moving files between OneDrive folders, sending Outlook email notifications — without additional Power Automate licences.

Premium Power Automate functionality requires either Power Automate Premium at $15 per user per month (previously Power Automate Per User) or Power Automate Process at $150 per bot per month (previously Power Automate Per Flow, designed for flows that run without a specific user context). Premium entitlements are required for flows using premium connectors, Dataverse access, process mining features, attended and unattended robotic process automation (RPA), and AI Builder actions within flows.

The same premium connector boundary applies. An automated flow that starts with a SharePoint trigger and posts to Teams can run on seeded entitlements. The same flow with an additional action that writes to Dataverse or calls a premium connector requires every user who owns or shares the flow to hold a Power Automate Premium licence. For enterprise environments with hundreds of citizen-developed automations, this boundary creates significant compliance exposure if not actively governed.

Dataverse: The Storage Cost That Escalates

Dataverse is Microsoft's structured data platform for Power Apps, Dynamics 365, and Copilot Studio. It provides a relational database with security roles, business rules, and API access — and it carries a per-GB storage cost that is one of the most consistently underestimated components of Power Platform total cost.

Every tenant receives a default Dataverse storage allocation: 10 GB of database storage and 20 GB of file storage at the tenant level, plus 250 MB of database and 2 GB of file storage per user with a qualifying Power Platform licence. Beyond these allocations, additional database storage costs $40 per GB per month, and additional file storage costs $2 per GB per month.

At enterprise scale, Dataverse accumulates data quickly. A moderate deployment with 1,000 Power Apps users generating workflow records, attachments, and audit logs can consume the default allocation within 12 to 18 months. At $40 per GB per month for database storage, a 100 GB overage costs $4,000 per month — $48,000 per year — before any additional file storage costs.

Organisations building Power Platform solutions should implement Dataverse storage governance from the first deployment: define data retention policies, archive inactive records to cheaper storage tiers, purge test and development data from production environments, and model storage growth projections before committing to Dataverse-backed application architectures at scale.

Power Pages: The Web Application Licensing Dimension

Power Pages is the external-facing component of Power Platform — a low-code platform for building authenticated web portals and public-facing websites powered by Dataverse. Power Pages licensing differs from the internal user model: it charges per authenticated session and per anonymous user visit rather than per named internal user.

The Power Pages Authenticated Users capacity add-on is priced at $200 per 100 authenticated users per month (roughly $2 per external user per month). The Anonymous Users add-on covers public-facing pages at $75 per 50,000 anonymous users per month. For organisations building customer portals, partner extranets, or external service delivery platforms on Power Pages, user volume projections are essential to understanding total cost — external user counts can scale dramatically faster than internal user counts, and per-session pricing creates cost variability that is difficult to budget without accurate usage forecasting.

Copilot Studio: Session-Based Pricing

Copilot Studio (formerly Power Virtual Agents) is the Power Platform component for building AI-powered conversational agents and bots. Its licensing model operates on sessions rather than named users. A session is defined as a conversation lasting up to 60 minutes or 100 turns of interaction. Copilot Studio is priced at $200 per 2,500 sessions per month — approximately $0.08 per session.

The session-based model creates predictability challenges for organisations with variable chatbot usage. A customer service bot that handles 10,000 sessions in a typical month and 50,000 during peak periods requires either an overprovision to the peak level or a variable billing arrangement. The cost difference between steady-state and peak — $800 versus $4,000 per month in this example — represents significant budget variability that internal IT cost models rarely capture accurately before deployment.

Copilot Studio sessions that use generative AI capabilities (Azure OpenAI-backed responses rather than rule-based flows) consume additional resources beyond the base session charge. Organisations deploying generative AI copilots should model session complexity alongside session volume, as AI-intensive sessions carry higher compute cost than standard rule-based dialogue sessions.

AI Builder Credits: The 2026 Change

Microsoft has announced the removal of complimentary AI Builder credits from Power Apps Premium and Power Automate Premium licences effective 1 November 2026. Currently, these licences include 5,000 AI Builder credits per month, which can be used for document processing (AI models that extract data from invoices, forms, and documents), prediction, object detection, and other AI Builder capabilities.

After November 2026, organisations that rely on AI Builder functionality within Power Apps or Power Automate flows will need to purchase AI Builder capacity add-ons separately. AI Builder add-on pricing is based on credits at $500 per 1 million credits per month. The practical impact depends on how extensively current deployments use AI Builder models — organisations with heavy document processing automation built on Power Automate should audit AI Builder usage and model the incremental cost of the add-on before the November 2026 deadline.

Power Platform in the M365 Licensing Context

Power Platform sits within the broader Microsoft licensing context in ways that affect procurement and negotiation. M365 E3, E5, and the new E7 — Microsoft's current top SKU above E5, bundling advanced AI and security capabilities — all include seeded Power Platform entitlements. The degree of seeded coverage differs: E5 and E7 include slightly enhanced entitlements compared with E3, but none eliminate the need for standalone Power Platform licences when premium connectors or Dataverse are required.

The EA negotiation implication is that Power Platform licence costs can be negotiated as a bundled component of the broader Microsoft EA renewal. Organisations that are simultaneously renewing M365, Azure, and Power Platform commitments should negotiate Power Platform as a package, not as isolated line items. EA-negotiated discounts on Power Apps Premium and Power Automate Premium can reach 15 to 20 percent below list price for large deployments incorporated into the EA at renewal, versus 0 to 5 percent for CSP purchases.

Six Governance Priorities

1. Map every application to its connector profile: Maintain a live registry of Power Apps and Power Automate flows, categorised by whether they use seeded-entitlement connectors or premium connectors. Any new connection to a premium source triggers a licence review before deployment.

2. Implement Dataverse storage monitoring from day one: Deploy automated alerts when Dataverse database storage exceeds 50 percent and 80 percent of allocation. Establish a data retention policy before overage costs accumulate.

3. Model Per App migration for affected deployments: If your organisation held Per App licences before January 2026 and relied on them for targeted deployments, model the Per App to Premium migration cost explicitly. The 4x price increase per user per app is material for large single-application user populations.

4. Audit AI Builder usage before November 2026: Run a report from the Power Platform admin centre on AI Builder credit consumption by environment. Quantify monthly credit usage and model the post-November 2026 cost under the add-on pricing structure.

5. Include Power Platform in EA negotiations: Do not procure Power Apps Premium or Power Automate Premium through CSP if an EA renewal is within 12 months. The EA negotiation window delivers meaningfully better commercial terms for large-scale Power Platform deployments.

6. Engage independent Microsoft EA advisory specialists: Power Platform licensing complexity — seeded boundaries, premium connector triggers, storage overages, session-based pricing, AI credit changes — requires specialist knowledge to govern effectively. An independent advisor identifies cost exposure before it materialises as a True-Up surprise.