This week, I had the privilege of attending the 2nd annual Microsoft Fabric Community Conference. For a product that's still in its infancy (~15 months), it was amazing to witness the excitement and enthusiasm from a very diverse community that spans people from literally all kinds of data backgrounds: SQL Server lifers, Power BI developers, data scientists, and even business analysts.
This year's event was jam packed with plenty of exciting announcements for new and upcoming features. It was a lot to take in, so I thought I'd do my best to organize my notes/thoughts in this recap post. There's no substitute for 3 days' worth of deep-dive sessions, but I'll also do my best to provide links to relevant resources for further investigation/study. And if you don't find what you're looking for, we're always happy to answer any questions you may have via phone/email/carrier pigeon.
State of the Union
These days, there's a ton of competition in the data and AI space, and it feels like disruption literally happens every day (hour?). So, it was very interesting to hear from Microsoft's perspective where they see the market, how Fabric is faring, what they're hearing from customers, and so forth.
By the Numbers
In terms of sheer growth, Fabric's numbers are off the charts:
Over 19K customers are now running Fabric
70% of the Fortune 500 are running Fabric and more than 50% run 3+ Fabric workloads
95% of the Fortune 500 are using Power BI
540K+ customers are running Power BI
There's an estimated 7M+ Power BI developers in the world
Microsoft was also proud to announce that they are literally 1-of-1 in terms of software vendors that are recognized as leaders in Gartner's key Magic Quadrants for data and AI:
Data Integration Tools
Cloud Databases
Cloud AI Developer Services
Analytics & BI Platforms
Figure 1: Microsoft's Unique Position in the Gartner Magic Quadrant(s) for Data & AI
What Customers Are Saying
While the general customer response I observed this week was overwhelmingly positive, there were several consistent themes in both terms of things customers were asking for and where Microsoft is already heading with the product roadmap(s). Let's break these down.
More Enterprise-Grade Hardening
Many of the announcements that drew the most excitement were related to enterprise-grade hardening. For this reporter, these features can be grouped into three distinct categories:
Features that arguably probably should have been there before, but are just now getting rolled out (e.g., in the next 30-90 days).
APIs/SDKs/CLI interfaces that can be used to develop more elaborate CI/CD pipelines.
Tools/features that both Microsoft and customers alike are discovering they need when running increasingly larger and more complex workloads in a shared, SaaS-style environment. In other words, items that are definitely needed, but no one could have predicted we'd need 15-18 months ago.
Feature Parity with Azure
For data engineers and data scientists, there's plenty of demand for feature parity with functionally equivalent tools/services in Azure. For example, there have been a lot of grumblings around the limitations of Data Factory in Fabric when compared to Azure Data Factory. As we'll see in a moment, many (most?) of these issues are getting addressed this upcoming quarter.
More Security & Governance Features
Security and governance were also big topics, and customers (rightfully) pointed to some of the holes in the end-to-end tapestry that is Fabric and Purview. Although there were several important announcements here, this was one area that Microsoft acknowledged as needing more work. The overall feedback from Microsoft product owners was that this would be a key focus for the next 12 months.
Announcements
Now let’s get into the good stuff. Below is a breakdown of the biggest Fabric announcements from the conference, organized by functional area to help you zero in on what matters most to your role. See this article from the Microsoft Fabric blog for further details on these announcements.
AI & Data Engineering
During the opening keynote, Arun Ulag, Corporate VP of Azure Data, made the statement that "Data and AI is a team sport", emphasizing the huge role that Fabric plays in democratizing data access for all user personas. Along those lines, there were some very exciting announcements related to Fabric and Azure AI Foundry.
Fabric Data Agents
As organizations work to build more capable, context-aware AI experiences, Fabric figures to become more and more of a go-to source for curated, trustworthy data. Fabric data agents—formerly known as AI skills—are poised to play a key role in this shift.
Fabric data agents do way more than just retrieve data from OneLake; they work to understand how the data is structured, what it represents, and when it's relevant, making enterprise data more accessible and usable for AI-driven solutions. With new integrations between Microsoft Fabric and Azure AI Foundry, developers can now bring these data agents into customized, conversational AI solutions. The following demo video illustrates how this works.
User Data Functions (Preview)
User Data Functions are a very interesting new feature introduced for developers that want to build reusable business logic on top of Fabric data. These functions are developed in Python and callable from data pipelines and notebooks. In some respects, you can think of them as being conceptually similar to Azure Functions in the sense that we can develop reusable functions that can be invoked from multiple contexts.
AI Functions (Preview)
AI Functions were originally introduced late last year. Now, we're getting more information about what kinds of problems they're intended to solve.
In a nutshell, AI Functions in Microsoft Fabric are intended to make it easy for you to enrich your data with large language model (LLM) capabilities right from within your data pipelines. Specifically, AI Functions allow you to apply natural language processing tasks like summarization, sentiment analysis, and classification without needing to write complex code or manage external AI services.
You can apply AI enrichment directly within Dataflows Gen2 and Pipelines, use predefined templates to get started quickly, and customize models with parameters and prompts tailored to your specific business needs. Similar to AI Builder in the Power Platform, AI Functions are positioned as a simple yet powerful way to bring generative AI into your everyday data workflows. See this announcement for more information on AI Functions.
Copilot Everywhere
As you might expect, there were several announcements around new Copilot features. However, the one that got the most excitement was the one where Microsoft announced that Copilot will now be accessible from any paid SKUs in Fabric. So, even if you're on an F2 SKU, you'll be able to leverage Copilot to help with various development efforts. You can read more about this exciting announcement here.
This veteran reporter was also excited to learn that you'll soon be able to access Copilot in Fabric Notebooks without having to first install any Python library packages.
Data Integration
Microsoft announced several notable enhancements related to data integration that are aimed at streamlining data movement and orchestration. Some notable items:
Gen2 dataflows are getting even more powerful with the general availability of Incremental Refresh and a new preview feature that allows you to convert Dataflow Gen1 to Gen2.
Database Mirroring now supports Azure Database for PostgreSQL flexible server and can connect to data sources over On-premises and Virtual Network Data Gateways.
Apache Airflow integration is now generally available, offering a serverless runtime to run DAGs directly in Fabric.
The new Copy job feature simplifies moving data between sources and supports both batch and incremental transfers.
And for those focused on orchestration, Fabric now includes preview capabilities for building metadata-driven pipelines and parameterized Dataflow Gen2 executions using Azure Key Vault and variable libraries.
Mark Kromer, the Cloud Data Analytics Product Leader at Microsoft, provides an excellent breakdown of all these new features in the Fabric blog here.
OneLake and Data Storage
Microsoft dropped a flurry of updates around data storage, all centered on making OneLake an even more open and flexible foundation for running your analytics workloads. From expanded support for mirroring to enhancements in how data is managed, secured, and shared, there’s a lot to unpack.
New Mirroring Options
In addition to the GA mirroring solutions around Snowflake and Azure SQL, Microsoft announced the public preview for mirroring from CosmosDB, Azure PostgreSQL, Azure SQL Managed Instances, and Azure Databricks Catalog. That's not to mention open mirroring solutions being developed by partners/ISVs as we described in our series on SAP & Fabric here.
Announcements were also made around mirroring data sources that sit behind a firewall. There's more work to be done, but Microsoft certainly seems committed to extending the reach of mirroring in Fabric to every corner of the enterprise (and beyond).
On a semi-related note, it was great to see the continued commitment towards open technologies like Apache Iceberg. At one point during the keynote, Chris Child, VP of Product Management at Snowflake, came on stage and talked about some of the investments Snowflake and Microsoft are making to make data from Snowflake & OneLake accessible without requiring replication.
OneLake Catalog
Last year, Microsoft introduced the OneLake Catalog as "a centralized place that helps you find, explore, and use the Fabric items you need, and govern the data you own". This was a great addition to the Fabric authoring experience and a one-stop shop for all types of Fabric workspace items.
Now, Microsoft is extending the reach of this catalog to tools like Excel and Power BI desktop. This is a good example of some small enhancements that streamline the Fabric experience across workspaces/personas.
Materialized Views for Lakehouses
This was a very welcomed addition to the developer tool bag. Besides the obvious performance benefits, it appears that these materialized views will be treated like first-class citizens in Fabric and not subject to the same limitations you see with traditional views which are only accessible via the T-SQL endpoint of a lakehouse.
Microsoft also demonstrated some advanced capabilities around data constraints. We're very excited to learn more about this one.
Migration Assistant for Azure Synapse Analytics
If you're already running production workloads on Azure Synapse, then Microsoft is making the path to Fabric smoother with the announcement of their new Migration Assistant tool. This tool makes it surprisingly easy to assess your existing environment and migrate workloads to Fabric with minimal disruption—making it easier to take advantage of Fabric’s unified data foundation whenever the time is right.
Real-Time Intelligence
While talking with a number of customers, there seems to be a noticeable awakening happening with Real-Time Intelligence (RTI). As teams get more hands-on with Fabric and start building confidence in the platform, RTI seems to be quickly shifting from an abstract concept to a real, tangible opportunity. Many customers are realizing that once their data is flowing into Fabric and the foundational pieces are in place, tapping into real-time insights isn’t just possible—it’s surprisingly within reach.
This week, most of the new announcements were focused on new streaming and event connectors (e.g., MQTT and Azure Event Grid). There were also announcements for the general availability of core RTI features like Fabric Events.
The announcement that generated the most excited was the introduction of the new digital twin builder which was briefly demonstrated during the keynote. There's not much information out there on this yet, but I'm sure we'll be hearing a lot more about this in the coming weeks. Meanwhile, you can find a more complete breakdown of new RTI announcements here.
Security & Governance
Security and governance might not be the flashiest topics at a data conference, but they’re table stakes for customers—and Microsoft clearly understands that. This year brought a wave of new features aimed at helping organizations stay in control as their data strategies (and AI use cases) grow more complex.
OneLake Security
A "one last thing"-style announcement that generated a lot of applause was the introduction of the new OneLake Security model. With OneLake Security, we can define access controls centrally within a data lakehouse, and those permissions are automatically enforced from everywhere—no matter how users access the data. Whether it’s through Power BI, T-SQL queries, notebooks, or even directly at the OneLake file system level, the same consistent security model applies.
What makes this especially appealing is how easy it is to implement. Authorizations can be defined using familiar T-SQL, which simplifies administration and makes it more approachable for teams already comfortable with SQL-based access controls. To see how it works, check out the video link below. You can also read more about it here.
Expanded Support for Service Principals
This was one of those enterprise-grade hardening topics that it sounds like a LOT of customers were asking for and, from what I can tell, Microsoft clearly answered. In the brief demo provided, it appears that support for service principals has been expanded pretty much across the board, working in similar fashion to the way service principals work in Azure. For those in the weeds with data integration with Data Factory, etc., this was a huge win.
Integration with Microsoft Purview
In our experience, customers love the promise of Purview, but aren't completely sold on Purview as a finished product yet. Recognizing this, it seems that the Fabric and Purview engineering team are hard at work developing some powerful new governance and security features to help organizations keep their data safe and compliant.
One of the big ones to watch is the upcoming preview of Microsoft Purview for Copilot in Power BI. This will make it easier to spot risks like sensitive data showing up in user prompts or responses, while also giving admins tools to investigate risky AI use and enforce policies through audit logs, retention rules, and insider risk management.
They’re also expanding Purview’s Data Loss Prevention (DLP) coverage to reach even more of the Fabric platform—including KQL databases and mirrored databases—so things like SSNs or other sensitive data can be detected and handled automatically, no matter where they pop up. And with the preview of Data Observability in the Unified Catalog, teams can now dig into data quality issues by seeing how data products are connected to different assets.
You can read more about the status of some of these developments in the Microsoft Purview blog.
DevOps
As we noted earlier, customers have been demanding more enterprise-grade DevOps support in Fabric—and Microsoft seems to be listening. There were a number of DevOps-related announcements that drew thunderous applause during the keynote.
A big highlight is the general availability of the Terraform provider for Fabric, allowing teams to fully define and deploy Fabric resources as code. Alongside that, a new Fabric command line interface was introduced, giving developers and DevOps teams the tools they need to automate deployments and streamline workflows.
Finally, there were quite a few CI/CD pipeline enhancements introduced that figure to bring Fabric development more in line with what you see elsewhere in Azure. Big wins all around.
Closing Thoughts
While it was an exhausting three days, it was really great to see how the Fabric community continues to grow as the product itself gains momentum. It's kind of crazy to imagine what the product might look like a year from now, but I think it's fair to say that Microsoft will continue to invest heavily in Fabric as it's the data foundation behind their larger AI play in the marketplace.
For us here at Bowdark, we're excited to not only see the product grow, but also continue to integrate more seamlessly with other parts of the Microsoft BizApps portfolio: Power Platform, Copilot, and even Dynamics 365. While the focus remains on driving business outcomes, it feels a little bit like Christmas getting to play with all these shiny new tech toys. :-)
