Microsoft Fabric in 2025: Complete Guide to New Features and Capabilities

Microsoft Fabric: The Data Revolution of 2025

Microsoft Fabric has established itself in 2025 as the reference unified analytics platform. With the announcement of major new features at FabCon 2025, Microsoft confirms its vision of a single platform to manage the entire data lifecycle. At Datakhi, we're already supporting several companies in their migration to Fabric.

Major Announcements of 2025

Copilot Now Accessible to Everyone

This is one of the most anticipated announcements: Fabric Copilot and AI capabilities are now available for all paid SKUs, and no longer just for F64 and higher capacities. This democratization, effective since late April 2025, allows SMBs to access the same AI tools as large enterprises.

Concretely, Copilot in Fabric allows you to:

• Generate DAX, SQL, or Python code in natural language
• Create Power BI visuals automatically
• Document data pipelines
• Perform guided exploratory analyses

Data Agents: Your Virtual Analysts

Fabric Data Agents represent a major advancement. These AI agents function as specialized virtual analysts, capable of:

• Reasoning about your data stored in OneLake
• Responding in real-time to business questions in natural language
• Integrating with Microsoft 365 Copilot for direct access from Teams
• Monitoring and alerting through autonomous "operations agents"

Data Agents automatically respect defined security permissions, ensuring that each user only accesses authorized data.

Osmos Acquisition: Intelligent Ingestion

In January 2026, Microsoft announced the acquisition of Osmos, a startup specializing in AI-powered data ingestion. This technology automatically transforms raw data (poorly formatted Excel files, inconsistent CSVs) into structured data ready for analysis.

Key Components of Fabric

OneLake: The Unified Data Lake

OneLake is the foundational building block of Microsoft Fabric. It's a unique and centralized Data Lake for the entire organization, inspired by the OneDrive concept but for analytical data. All Fabric data is automatically stored in OneLake in Delta Lake format (Parquet), which guarantees interoperability and performance.

Why it's revolutionary: no more data silos! Whether you use Power BI, Spark, or SQL, all your data is in the same place, with a single permission system. In 2025, OneLake is enhanced with:

• Centralized Row-Level and Column-Level controls: define your security rules once, they automatically apply everywhere (Spark, SQL, Power BI). No more duplicating permissions in each tool.
• OneLake Catalog: an integrated data catalog to manage metadata, understand lineage (where your data comes from, where it goes) and facilitate data discovery by teams.
• Enhanced Purview integration: automatic classification of sensitive data (PII, financial data) and application of sensitivity labels for GDPR compliance.
• Shortcuts: create shortcuts to external data (Azure Data Lake, AWS S3, Google Cloud Storage) without moving it, for facilitated data mesh architecture.

Real-Time Intelligence

Microsoft Fabric was recognized as Leader in the Forrester Wave Streaming Data Platforms Q4 2025. This component allows ingesting, processing, and analyzing data in real-time, ideal for IoT, fraud detection, or application monitoring.

How it works: streaming data arrives via Event Streams (Kafka-compatible), is stored in Eventhouse (a real-time optimized database based on Kusto/KQL), then can trigger automatic actions via Activator.

2025 updates include:

• Activator: up to 10,000 events per second (10x vs 2024). Create automatic alerts and actions based on your real-time data without coding.
• Maps in Fabric: native geospatial visualization to display your data on interactive maps.
• ArcGIS integration: advanced Esri cartography directly in your reports, at no additional license cost.

Enhanced Data Engineering

Fabric's Data Engineering component offers a complete environment for data engineers, with Lakehouses (Data Lake + Data Warehouse combination), notebooks, and data pipelines.

• Fabric Runtime 2.0: access to Spark 4.0 and Delta Lake 4.0 for increased performance and new features (UniForm for Iceberg compatibility).
• Native Python Notebooks: integrated Jupyter environment with multi-kernel support (Python, Scala, R) and pre-installed data science libraries (pandas, scikit-learn, etc.).
• T-SQL in notebooks: combine SQL and Python in the same notebook for maximum flexibility.
• LLM Transformations: native functions for translation, entity extraction, and sentiment analysis directly in your pipelines.

Data Warehouse

The Fabric Data Warehouse is a serverless SQL data warehouse, 100% T-SQL compatible. Unlike traditional solutions, you don't manage any infrastructure: Microsoft handles automatic scaling.

Key points: standard SQL queries, native integration with Power BI (ultra-performant Direct Lake mode), and storage in OneLake in Delta format for total interoperability with other components.

Power BI: Even More Powerful

Power BI remains the reference visualization and BI tool, now fully integrated into Fabric. Power BI datasets become "Semantic Models" accessible by all components.

• Write-back: modify data directly from your reports (budgeting, collaborative planning).
• Multiple organizational applications per workspace: publish multiple business applications from the same workspace.
• Contextualized Copilot: AI understands your business vocabulary and generates visuals, DAX measures, or insights in natural language.
• Direct Lake mode: Import mode performance with Direct Query freshness, the best of both worlds.

Governance and Security: The 2025 Priority

Microsoft insists: "AI is only safe if it's secured". With massive adoption of generative AI, data governance becomes critical. Microsoft's investments in 2025 are massive:

Data Observability

Real-time data quality monitoring allows detecting anomalies before they impact your analyses. Fabric automatically monitors data freshness, volume discrepancies, and pipeline errors, with configurable alerts.

Integrated Microsoft Purview

Purview is Microsoft's governance hub, now deeply integrated with Fabric:

• Automatic classification: AI automatically identifies sensitive data (credit card numbers, emails, health data) in your Lakehouses and Warehouses.
• Sensitivity labels: apply labels (Confidential, Public, Restricted) that follow data wherever it goes.
• DLP (Data Loss Prevention): block export or sharing of sensitive data according to your company policies.
• Data lineage: visualize where your data comes from and where it's used, essential for compliance and audit.

Domain Organization

Fabric allows organizing your data by Domains and Sub-domains corresponding to your businesses (Finance, HR, Marketing...). Each domain can have its own governance rules, Data Owners, and access policies, while remaining in a unified platform. This is the realization of Data Mesh in Fabric.

Enhanced Security

• Managed Private Endpoints: connect to your data sources securely via private links, without exposing your data on the Internet.
• Customer-managed keys (CMK): encrypt your data with your own keys for total control.
• Audit logs: track all actions performed in Fabric for regulatory compliance.

Pricing and Licenses

Microsoft Fabric uses a Capacity Units (CU) model with usage-based billing. The main options:

Capacity	CU/hour	Use case
F2	2 CU	POC, small teams
F16	16 CU	SMBs, departments
F64	64 CU	Enterprises (all features)
F128+	128+ CU	Large groups, intensive workloads

Since April 2025, Copilot is included in all paid capacities.

Optimize Your Costs with Datakhi

Datakhi is a Microsoft CSP (Cloud Solution Provider) partner, which means we are official resellers of Microsoft licenses. Concretely, you benefit from:

• Preferential rates: through Datakhi, your Fabric licenses (and Microsoft 365, Azure, Power BI Pro/Premium) are cheaper than direct
• Unified billing: a single invoice in euros, with a single point of contact
• Included support: our experts help you size the right capacity and optimize your costs
• Flexibility: adjust your capacity each month according to your real needs

Contact us for a personalized quote and discover how much you can save on your Microsoft licenses.

How to Migrate to Microsoft Fabric?

At Datakhi, we recommend a 4-phase approach:

1. Audit: analysis of your current architecture (Azure Synapse, Databricks, Power BI Service)
2. Strategy: definition of scope, priorities, and planning
3. Migration: progressive transfer of workloads with non-regression tests
4. Optimization: performance tuning and team training

Conclusion: Why Switch to Fabric in 2025?

Microsoft Fabric is no longer an option, it's a strategic necessity for data-driven companies. With the democratization of Copilot, the arrival of Data Agents, and strengthened governance, 2025 marks a turning point.

Ready to modernize your data stack? Contact our Fabric experts for tailor-made support.