Every Fabric Planning update announced at FabCon Europe in Barcelona, with a short demo video for each feature.
At FabCon Europe 2026, Microsoft announced a new set of capabilities for Planning in Microsoft Fabric. The updates make planning simpler to set up, faster at enterprise scale, more tightly connected to Power BI semantic models and the rest of Fabric, and easier to govern as adoption grows.
Microsoft's announcement covers the headline features: Planning in Microsoft Fabric: From insight to action. This post goes through the full list of updates, shows each one in a demo video, and marks which features are in preview or coming soon.
What's new with Planning
The Planning updates are grouped into five areas, followed by the limitations removed in this release.
1. Start faster
These updates cut down the setup needed before a team can begin working on the numbers.
No-setup connectivity for semantic models
Connecting a plan to a semantic model is now as simple as building a Power BI report: a single click, with the same workspace permissions you already use in Power BI.
Create Plan → choose semantic model → connect → start planning.
Until now, most onboarding effort went into setting up and maintaining the DMTS connection, and into fixing plans when that connection expired. That setup is gone.
- No DMTS connection. There is no DMTS connection to create or share, and no DMTS connection expiry to recover from.
- No shared cloud connection or service principal. Planners connect with their own signed-in identity.
- No tenant settings to configure. The four tenant settings Fabric Planning relies on, including the XMLA endpoint settings, are on by default. You only need to turn one on if it was turned off in your tenant.
- No additional permissions. Workspace Admin or Member rights are not required, and Contributors do not need Build permission on the semantic model. If you can build a Power BI report on the model, you can plan on it.
- Security enforced per user. Direct Lake and DirectQuery run natively over SSO, and row-level and object-level security from the semantic model apply automatically to each signed-in user.
| Before | Now |
| Create and share a DMTS connection | Not needed |
| Recover plans after a DMTS connection expires | Not needed |
| Configure a shared cloud connection and service principal | Not needed |
| Turn on tenant settings, including XMLA | On by default |
| Workspace Admin or Member rights, and Build permission on the semantic model | Same permissions as building a Power BI report |
Existing permissions and data access controls continue to apply, so nothing new has to be secured.
Home Layout
Different teams often work inside the same Plan item but need different information. Home Layout lets you create dedicated landing pages for each role within one Plan item.
- Role-based pages built from hero banners, sections, images and sheet cards.
- Respects role security, so hidden sheets never appear.
- Layout priority, plus an item baseline default.
Each team lands on what it needs first, with less navigation, and one Plan item can serve every team.
2. Plan at scale
These updates support detailed, driver-based plans that stay connected to enterprise targets.
Native Planning Engine Preview
The new Native Planning Engine is built on Rust and Apache Arrow with tight OneLake integration. It delivers up to 10x improvements in performance and scale.
- Work with up to 12 million cells in a planning model, with dependent calculations updating as inputs change.
- Input and allocate across large, detailed business models.
- Any number of dimensions, measures or scenarios, with six model options at scale.
- Keep departmental (line of business) plans connected across the enterprise.
How much scale do you need? Recommendations
Semantic models can hold billions of rows, but a plan usually works with a few dimensions in rows, time in columns and a set of measures. Most plans need no more than 1.2 million cells. With filters and row-level security (RLS) on the semantic model, that is enough scale for most customers.
The Native Planning Engine supports up to 12 million cells in a planning model. Large plans are still limited by two things outside the engine:
- Fetch time. Loading a large volume of data from the semantic model takes time.
- Power BI capacity limits. Power BI limits how much data a query can fetch and how many queries can run at once. When many users pull large volumes of data at the same time, the semantic model's Analysis Services engine can throttle their queries.
Our recommendations:
- Limit what each user fetches. Use filters and RLS so most users load only the data they plan on. Typically only a few planning power users need the full data set, for example to allocate at the grand-total level.
- Choose a modeling method built for the grain. When a plan needs this many cells, evaluate:
- PowerTable integration for planning at line-item grain, such as SKUs, employees or customers.
- Cube for multi-dimensional planning.
- Match the method to the use case. Fabric Planning supports several modeling methods for bottom-up operational planning, top-down financial planning, driver-based planning, scenario planning and strategic planning. Follow the implementation best practices for your use case.
Multi-driver, multi-measure simulations
You can now test several drivers together and see their combined effect on the measures that matter.
- Explore price, volume, cost and staffing assumptions together. For example, a sales plan might assume higher volume and a lower selling price, and operations can then test the material, labor and overtime implications.
- In Matrix or Tree View, changes flow through the configured driver relationships and hierarchy detail through to business KPIs such as profit.
- Create scenarios such as Best Case and compare them before committing to the forecast.
More ways to model and deliver planning
- Custom fiscal year planning supports calendars whose periods do not begin in January.
- Stepped, Outline and Tree layouts let teams shape the planning grid around how their business works.
- Measure modeling builds plans from the measures already in your semantic model.
- Advanced Lookup brings XLookup into the planning grid and enables PowerTable lookups into Fabric SQL.
3. Put plans to work
These updates move approved plans and forecasts into analysis, agents and downstream workflows.
Semantic model integration
Planners can now write plan data back into the same semantic model the plan is built on, so actuals and plans live in one governed model.
- After you configure the writeback destination, add the resulting table and its relationships to your existing semantic model directly from the planning UI.
- The writeback data is then available to Power BI reporting, Copilot and Data Agents without a separate manual modeling step.
- Compare actuals and forecasts in Power BI.
Fabric IQ ontology and Data Agents
Once plan data is in the connected semantic model, Fabric IQ can use plans and forecasts alongside actuals.
- Connect planning results to the ontology, so targets, forecasts and assumptions are represented as business entities, properties and relationships.
- Give Data Agents and configured Copilot experiences plan context.
- Reason over both historical results and forward-looking forecasts, for example when analyzing variance or evaluating scenarios.
Event Triggers
A plan rarely stops when a value is submitted. Event Triggers let planners configure, from the planning UI, what runs when a writeback starts, succeeds or fails.
- Run Fabric pipelines, notebooks, Dataflow Gen2, semantic model refreshes and SQL stored procedures.
- Send automated Microsoft Teams notifications and Outlook email.
- Pass writeback details and execution metadata directly as action parameters.
- Configure multiple triggers per planning sheet for different automation paths.
Execute Stored Procedure in PowerTable Automations
PowerTable Automations already provide no-code actions for creating, updating and managing records. Some processes depend on logic that lives in the database. Execute Stored Procedure brings that logic into the same automated workflow.
- Run stored procedures from your Fabric SQL database as an action in PowerTable Automations.
- Select procedures by schema from a dropdown, with no integration code or external scheduling.
- Pass values from the triggering record, such as a customer or order key, so each run targets the right data.
- Combine with no-code actions such as Create, Update and Find Records.
- Reuse the procedures your data team already maintains for aggregations, recalculations and multi-table updates.
OneLake writeback (Coming soon)
Fabric Planning will add OneLake as a native writeback destination, as an alternative to Fabric SQL Database.
- Store planning inputs, budgets and forecasts as native Delta Parquet tables in a Lakehouse.
- Reuse the data through OneLake shortcuts and compatible Fabric workloads.
- Remove the need for a separate SQL writeback database for teams with a lake-first architecture.
- Scale: with OneLake as a direct writeback destination, Fabric Planning will support writeback of 10M+ cells.
4. Reach the business
These updates bring planning into the apps people already use and extend it to the people who know the business.
Planning in organizational apps
Planning Sheets and Intelligence Sheet reports can now be embedded as pages in a Power BI or Fabric organizational app, alongside Power BI reports and related analytics.
- Users open one application link to review performance, reach the plan and continue the workflow.
- Navigation and content can be curated for each business audience.
- Existing app audiences and security rules govern access. No separate share link or direct workspace access is needed.
- Org apps with Plan content render the same way in Microsoft Teams and SharePoint.
Planning and PowerTable in the Intelligence Sheet
A planner who builds a detailed planning sheet often also needs KPIs, charts and a management-ready view of the same plan. The Intelligence Sheet can now hold Planning and PowerTable content together in one sheet.
- Update plans live and see the impact across KPIs and visuals.
- Embed PowerTable grids and Forms alongside Gantt, Resource, Kanban and Calendar views, all inheriting the report's theme.
- Combine plan data with operational detail to investigate variances and drivers.
Direct sharing of Plan items
A Plan item can now be shared directly with any user, even if they are not a member or viewer of the workspace it lives in.
- The item owner grants the user read or edit access to the semantic model the plan is built on.
- Users open the plan in read mode and can still take Stakeholder actions, such as entering data in input columns and adding comments.
- Participation stays aligned with role and data permissions.
Microsoft Entra B2B guest access
Guest users invited through Microsoft Entra B2B can now access and collaborate on Plan items in the same way as internal users.
- Stop provisioning internal accounts for external partners.
- Standard B2B invitation and access management applies, so governance stays central.
- Supports planning across organizational boundaries: joint ventures, external finance teams, consulting engagements and group consolidations.
Notes, Marker Mode and Annotation
Report authors can add narrative and context directly around the data, so a report explains itself when it is presented.
- Notes for live editing, to tell the story around the numbers.
- Marker Mode to draw attention to key takeaways and data points.
- Annotation to connect notes from the semantic model directly to a visual.
IBCS® certified reporting
Fabric Planning is IBCS® certified, aligned with ISO 24896:2026.
- Full semantic coverage and 100+ charts and templates (C01–C13, T01–T05).
- One-click notation, with automatic AC, PL, PY and FC scenarios.
- IBCS-aligned advanced visuals such as Marimekko, box plots and bullet charts.
- Uniform notation, scaling and color across multiple reports through themes.
5. Govern growth
These updates give administrators control over access, capacity and permissions as more teams adopt planning.
Tenant-level session upgrade controls
Until now, any user who entered data, wrote back changes or collaborated was upgraded automatically from a Viewer session to a Stakeholder session. Three new Plan tenant settings in the Fabric Admin Portal let administrators decide who can upgrade and see capacity impact before it happens.
- Control who can upgrade to a Planner session and who can upgrade to a Stakeholder session. Access defaults to the entire organization and can be restricted to specific security groups.
- Turn on an oversubscription warning that appears before a session upgrade is likely to exceed capacity, for example a second concurrent Planner session on an F2.
- Override settings for individual capacities through Delegated Tenant Settings.
- Users outside the allowed groups can still view content in Reading view.
Capacity throttling
Fabric capacity controls now apply consistently to planning as it reaches a wider audience.
- New sessions and session upgrades are blocked when a capacity is throttled.
- Additional demand is contained during periods of capacity pressure.
- Administrators get consistent workload operating controls.
Row-level security for PowerTable
Teams that write back to shared tables rarely need the same access to the same data. Row-level security brings data-driven access control into PowerTable, down to the individual row.
- Control who can read, update, delete or insert each row, based on values such as territory or region.
- Combine rules into policies and policies into roles, then assign multiple roles to any user.
- Use user attributes such as email or manager, so access follows each person automatically.
- Serve every team from one shared sheet instead of duplicating sheets for each audience.
Contributor security controls Coming soon
A new setting will let administrators control which Contributors can change security and governance settings in a Plan item.
Private Link support Preview
Fabric Planning supports Microsoft Fabric Private Link at both tenant and workspace level.
- Route planning traffic through the Microsoft private network.
- Restrict public access so Plan items, semantic model connections and writebacks stay within your network boundary.
- Meets network-isolation requirements in regulated industries such as financial services, healthcare and government.
PowerTable-only billing
Microsoft recently announced a billing change for non-planning editors. For PowerTable-only data management, editors are billed as Stakeholders and consumers as Viewers, so teams get full data management without a Planner session.
6. From known limitations to known wins
Several limitations from earlier releases have now been removed.
Access and connectivity
| Feature | March 2026 | Now |
| B2B user support | No | Yes |
| Private Link support | No | Yes (Preview) |
| Semantic model connection authentication | OAuth / service principal | All supported authentication methods |
| Admin/Build permission required on semantic model | Required | No |
| Direct Lake semantic models need extra configuration | Yes | No |
Scale and capacity
| Feature | March 2026 | Now |
| Power BI Pro / PPU / lower-capacity SKUs (no XMLA support) | No | Yes |
| Bulk data input limit | 1M rows | 10M rows |
Workspace, deployment and data security
| Feature | March 2026 | Now |
| CI/CD deployment with a service principal | No | Yes |
| Database-level row-level security (RLS) | No | Yes |
| DMTS connection recovery | Manual rebuild needed | Yes |
| Contributor role: create/share cloud connections | No | Not applicable |
| Contributor role: Plan items requiring embed tokens | No | Not applicable |
What comes next
Microsoft's stated next step is an API surface for the planning lifecycle, so an agent can work inside a plan the way a planner does: draft a forecast, run a simulation across drivers, compare scenarios before a commitment is made, write the result back, and start the downstream work it sets in motion.
What's New with Fabric Planning: webinar
We will walk through these updates in a live session, with demos and time for questions: Register now

Nice work!
Wow
With the upcoming native planning engine, will the writeback destination still be Fabric SQL database or did I hear/read somewhere that there will be writeback to other Fabric databases (e.g. Warehouse)?