FabCon 2025 was by far the most full schedule I have ever kept at a conference. Yet, I have never come to the conclusion of a conference feeling more invigorated, more inspired, or more grateful than this one. My intent was to post a daily recap throughout the conference. The days were so wonderfully packed that when I had time to write, what I needed more was rest to prepare for the next day. Much of the “extra” in my schedule this year was time spent with colleagues turned friends. Eating tacos, discussing highlights from the day, and comparing session notes. In times past, I would have had dinner alone but this year I was intentional to connect more. To engage the community more than I ever have. After all, FabCon’s full name is Microsoft Fabric Community Conference. And this week it was really the people who energized me. So the biggest takeaway from the week was the community.

There also happened to be ALOT of announcements in Fabric and Power BI this week. Read about all of the announcements from the keynote here, but here a a few that stuck out to me:

Copilot will be available for ALL paid Fabric SKUs

I think this could be the biggest announcement of the week. This means that anyone who purchases a Fabric SKU, even as low as an F2, will have access to Copilot and Fabric Data Agents. I work for an organization that is leveraging Fabric, but we haven’t scaled to an F64 capacity yet (before now an F64 SKU was needed for Copilot access in Fabric), but I am excited to have access to more features now. Fabric Data Agents has a lot of potential. The ability to connect and train Copilot on data in my organization’s OneLake, and in particular the semantic models that I build, is intriguing. I plan to post more about this as I dig in and get hands on.

Calendars in DAX

This announcement didn’t make the keynote, but was made in Marco Russo and Jay ter Heerdt’s DAX breakout session. The feature is not available yet, but will introduce the concept of “calendars” in a date table. What it means for Power BI developers is that we will now be able to use the built in date functions across financial calendars, ISO calendars, and whatever other calendar type requirements may throw at us! I’ll be sad to see the old “Greatest Formula in the World” pattern come to an end, but writing time intelligence will be so much easier!

Composite Models: Direct Lake and Import Mode

I anticipate using Direct Lake more and more in my development work, so having a composite model option available will be handy. Composite model availability means that developers have a work around for features that are currently lacking in Direct Lake mode, namely calculated columns. In a composite model, tables requiring calculated columns could be stored in import mode alongside other tables in Direct Lake mode. Marco Russo’s note on the relationships between these tables is noteworthy also. More on that at the link above.

Variable Libraries

I have wanted the ability to define variables once and reuse them across a Fabric tenant since writing my first notebook. There will certainly be a lot of use cases for Variable libraries, but I’m immediately looking forward to the efficiency of reusing variables.

User Data Functions

UDFs actually got announced for both Fabric and DAX. More efficiency!

Fabric UDFs (in Preview at the time of this writing) will allow developers to write functions that can be called in Fabric notebooks, pipelines, or even in a Power BI report. This means that custom algorithms or business logic can be written once, stored in a function, then invoked across the platform. The addition of UDFs in Fabric feels like a natural progression of the platform. During a session on this topic, Luis Bosquez and Sunitha Muthukrishna from Microsoft demoed how a Fabric UDF can be called by a button visual in Power BI, pulling in values from input slicers. I’ve got a use case for this in mind! So here’s another announcement that I’ll plan to post more on as I get hands on!

DAX UDFs bring the same concept into the semantic model. DAX UDFs has the potential to change how we write DAX, much like DAX variables did.

Build Direct Lake Semantic Models within Power BI Desktop

Building semantic models in the browser through the Fabric UI hasn’t been terrible, but I like the convenience of working in the desktop tool that I’m already used to modelling in. The real benefit I see here is that during report development, we have the ability to adjust and enhance the model without jumping to another window.

Other Learnings & Tidbits

• The reusability of TMDL – easy to copy snippets of the scripts between semantic models
• Semantic Link and Semantic Link labs are keys to furthering my Fabric development
• Charty Party is a fun game. Even more fun when you play with a bunch of data nerds. (I am not affiliated with the game or Amazon in any way)
• Tacos El Gordo is a fantastic Tijuana style taco restaurant. Close enough to the Vegas strip to be convenient but far enough away to escape the party crowd. Thanks for the recommendation Chuy Varella.

In Conclusion

What did you learn? What new features are you most excited about? Drop your comments below. Who else is ready for FabCon 2026?!?! Hopefully I’ll see you there!