15/04/2026

FabCon 2026 Highlights: Microsoft Fabric gets more developer-friendly and production-ready

I was not in Atlanta myself but between the March 2026 feature summary, the side announcements and the stream of posts on LinkedIn, one thing became pretty clear: Microsoft Fabric is starting to look less like a platform full of promising ideas and more like one you can seriously build and operate on. That may sound less exciting than another flashy AI demo. But honestly, this is the part that matters.

The announcements around FabCon 2026 were not just about adding “more features”. A lot of them were about making Fabric more usable for real teams. In this blog, I want to focus on the updates that stood out to me most. Not necessarily the loudest ones, but the ones that feel meaningful if you are actually working with Fabric as a data platform.

Fabric is leaning further into real software engineering

One of the clearest themes at FabCon 2026 was that Fabric is becoming friendlier to teams that build like software engineers rather than portal-only users. Branched workspaces, now in preview, bring a more explicit branch-out experience for feature workspaces. Selective branching makes that even more practical because teams can include only the items they actually need instead of cloning an entire workspace and carrying unnecessary items into every feature branch.

Figure 1. Selective Branching Source: Microsoft

Fabric now displays compare code changes before syncing with Git. That sounds small but it removes a very real source of anxiety in collaborative work. When developers can inspect item-level and file-level diffs before they commit, update or undo.

Microsoft also kept pushing the code-first story beyond Git alone. Notebook auto-binding in Git reduces environment-specific reconfiguration when notebooks move across workspaces. Support for notebook resource folders means helper modules, config files and small assets can move along with the notebook project as a single source-controlled unit. On top of that, the new deploy command in Fabric CLI makes automated deployments more natural. The Fabric CLI continues to improve and it can help you and your team move faster by simplifying automations.

Then there is Fabric MCP. The ability for GitHub Copilot, Codex and other AI tools to interact with Fabric through local or remote MCP servers is still early but strategically it may be one of the most important announcements. It points to a future where development, administration and support tasks in Fabric become increasingly agent-assisted. For teams already standardizing on Git, CI/CD and developer tooling, this is a strong signal that Fabric wants to meet them where they already work.

Figure 2. Fabric now shows item-level and file-level diffs before syncing changes with Git. Source: Microsoft.

Data engineering gets faster and more predictable

Data engineering announcements at FabCon 2026 were equally practical. Fabric Runtime 2.0 brings Spark 4.0, Delta Lake 4.0 and Python 3.12 into the platform. That alone makes the runtime worth watching because runtime version upgrades have a direct impact on library compatibility, performance and developer productivity.

Microsoft is also adding more explicit control over how Spark workloads behave in production. Custom Live Pools allow teams to keep dedicated Spark pools warm for recurring workloads, reducing cold-start delays and making performance more predictable. For frequently executed notebooks and production pipelines, that can remove one of the most frustrating sources of latency.

Resource Profile Configurations are another welcome step. Instead of tuning Spark settings ad hoc for every workload, teams can use reusable workload profiles such as read-heavy or write-heavy configurations. This is the kind of feature that reduces operational inconsistency and helps less specialized teams make better performance decisions without turning every project into a tuning exercise.

Figure 3. Live Pools for Custom Spark configurations

Materialized lake views also became more flexible with support for multiple schedules and PySpark in addition to SQL. That matters because it expands the feature beyond SQL-centric use cases and gives Spark-oriented teams more freedom to operationalize transformations in a structured way.

Figure 4. Execution of Materialized Lake Views

Warehousing and pipelines gain the operational features teams keep asking for

In the warehouse space, recovery options, including deleted warehouse recovery, improve safety nets for operational teams. Custom SQL pools introduce a more deliberate way to allocate compute across different workload types such as ad hoc analytics, reporting and pipelines. SQL audit logs are now generally available, which is especially relevant for regulated environments or teams that need stronger visibility into access patterns.

The general availability of reusable data sources for file-based SQL access is another strong addition. Defining a storage location once and reusing it by name makes queries cleaner and easier to maintain, especially in environments where OPENROWSET patterns can otherwise become repetitive and fragile.

On the pipeline side, the March update brought several features that improve day-to-day operations. Failure notifications help teams stop relying on manual monitoring for recurring runs. The ‘Lakehouse Maintenance’ activity brings optimize, V-Order and vacuum operations into pipeline orchestration. The ‘Refresh SQL endpoint’ activity helps keep SQL metadata aligned with upstream lakehouse changes. And the ‘SSIS Pipeline’ Activity offers a pragmatic migration bridge for organizations that still carry significant SSIS investments. However, it might be worth exploring migrating to cloud-native workflows instead of dragging along technical debt which may be incurred in SSIS packages.

Figure 5. Lakehouse Maintenance Activity in Fabric Data Pipeline.

Figure 6. Refresh SQL Endpoint metadate for Lakehouse in Fabric Data Pipeline

Figure 7. Invoke SSIS packages directly from Microsoft Fabric Source: Microsoft

This is why the March 2026 wave feels more mature than flashy. Microsoft is not only expanding what Fabric can do. It is also investing in how reliably teams can run, maintain, observe and migrate workloads once the platform becomes business-critical.

Figure 8. Custom SQL pools bring more deliberate workload allocation to Fabric Warehouse. Source: Microsoft

Mirroring keeps expanding the Fabric landing zone

Mirroring was another area where FabCon 2026 showed clear momentum. Support for SAP and Oracle databases moving to general availability is significant because it strengthens Fabric’s position in enterprise integration scenarios, where these systems remain central. Preview support for Azure Database for MySQL and SharePoint Lists broadens that story further, extending low-friction replication to both operational databases and business-managed data sources.

No single mirroring announcement may have been the loudest headline on its own but together they reinforce one of Fabric’s strongest value propositions: reducing the effort required to land source data in OneLake in an analytics-ready format. The less time teams spend building bespoke ingestion and replication patterns, the faster they can focus on modelling, governance and business value.

Closing

If I had to summarize FabCon 2026 in one sentence, it would be this: Microsoft Fabric is becoming easier to engineer, easier to operate and easier to connect. The March announcements did not just add new surface area. They reduced friction across the lifecycle of real data products.

For teams building serious Fabric workloads, several of these features are worth piloting quickly: branched workspaces and Git diffing for safer collaboration, Runtime 2.0 and resource profiles for more predictable Spark execution, new maintenance and refresh activities for stronger orchestration, and mirroring expansions for faster source onboarding. If Microsoft keeps executing in this direction, Fabric will not only remain a compelling unified data platform; it will become a far more comfortable platform to build on at scale.

Further reading

Fabric March 2026 Feature Summary
What’s new in Microsoft Fabric
Manage branches in Microsoft Fabric workspaces
Fabric Runtime 2.0 Experimental (Preview)
Workload management in Fabric Data Warehouse
What is Mirroring in Fabric?

Niels De Swaef

Consultant @ Lytix

Niels De Swaef is a Data Engineer passionate about using technology to solve complex problems and improve people’s lives. He has experience with the Microsoft Azure data stack, including Synapse, Databricks, Azure Data Factory, and CI/CD pipelines. Niels is also a lifelong learner who is always looking for new ways to use data to make a positive impact.