If you know Betteridge’s Law of Headlines, then you know the answer is no. But let’s get into it anyway.

Recently there was LinkedIn post that made a bunch of great and valid points but ended on an odd one.

Number one change would be removing Power BI from Fabric completely and doubling down on making it even easier for the average business user, as I have previously covered in some posts.

It’s hard for me to take this as a serious proposal instead of wishful thinking, but I think the author is being serious, so let’s treat it as such.

Historically, Microsoft has failed to stick the landing on big data

If you look back at the family tree of Microsoft Fabric, it’s a series of attempts to turn SQL Server into MPP and Big Data tools. None of which, as far as I can tell, ever gained significant popularity. Each time, the architecture would change, pivoting to the current hotness (MPP -> Hadoop -> Kubernetes -> Spark -> Databricks). Below are all tools that either died out or morphed their way into Fabric today.

• (2010). Parallel Data Warehouses. A MPP tool by DataAllegro that was tied to a HP Hardware Appliance. Never once did I hear about someone implementing this.
• (2014) Analytics Platform System. A rename and enhancement of PDW, adding in HDInsight. Never once did I hear about someone implementing this. Support ends in 2026.
• (2015) Azure SQL Data Warehouse. A migration of APS to the cloud, providing the ability to charge storage and compute separately. Positioned as a competitor to Redshift. I may have rarely heard of people using this, but nothing sticks out.
• (2019). Big Data Clusters. An overly complicated attempt to run a cluster of SQL Server nodes on Linux, supporting HDFS and Spark. It was killed off 3 years later.
• (2019) Azure Synapse Dedicated Pools. This was a new paint of coat Azure SQL Data Warehouse, put under the same umbrella as other products. I have in fact heard of some people using this. I found it incredibly frustrating to learn.
• (2023) Microsoft Fabric. Yet another evolution, replacing Synapse. Synapse is still supported but I haven’t seen any feature updates, so I would treat it as on life support.

That’s 6 products in 13 years. A new product every 2 years. If you are familiar with this saga, I can’t blame you for being pessimistic about the future of Fabric. Microsoft does not have a proven track record here.

Fabric would fail without Power BI

So is Fabric a distraction? Certainly. Should Power BI just be sliced off from Fabric, so it can continue to be a self-service B2C tool, and get the attention it deserves? Hell, no.

In my opinion, making such a suggestion completely misses the point. Fabric will fail without Power BI, full stop. Splitting would mean throwing in the towel for Microsoft and be highly embarrassing.

The only reason I have any faith in Fabric is because of Power BI and the amazing people who built Power BI. The only reason I have any confidence in Fabric is because of the proven pricing and development model of Power BI. The only reason I’m learning Fabric is because the fate of the two is inextricably bound now. I’m not doing it because I want to. We are all along for the ride whether we like it or not.

I have spent the past decade of my career successfully dodging Azure. I have never had to use Azure in any of my work, outside of very basic VMs for testing purposes. I have never learned how to use ADF, Azure SQL, Synapse, or any of that stuff. But that streak has ended with Fabric.

My customers are asking me about Fabric. I had to give a 5 day Power BI training, with one of the days on Fabric. Change is coming for us Power BI folks and I think consultants like me are mad that Microsoft moved our cheese. I get it. I spent a decade peacefully ignorant of what a lakehouse was until now, blah.

Is Power BI at risk? Of course it is! Microsoft Fabric is a massively ambitious project and a lot of development energy is going into adding new tools to Fabric like SQL DBs as well quality of life improvements. It’s a big bet and I estimate it will be another 2-3 years until it feels fully baked, just like it took Power BI 4 years. It’s a real concern right now.

Lastly, the logistics of detachment would be so complex and painful to MSFT that suggesting it is woefully naive. Many of the core PBI staff were moved to the Synapse side years ago. It’s a joint Fabric CAT team now.

Is MSFT supposed undo the deprecation of the P1 SKU and say “whoopsie-daisy”? “Hey sorry we scared you into signing a multi-year Fabric agreement, you can have your P1 back”? Seriously?

No, Odysseus has been tied to the mast. Fabric and Power BI sink or swim together. And for Power BI consultants like me, our careers sink or swim with it. Scary stuff!

Where Microsoft can do better

Currently I think there is a lot of room for improvement in the storytelling around which product to use when. I think there is room for improvement from massive tables and long user scenarios. I would love to see videos with clear do’s and don’ts, but I expect those will have to come from the community. I see a lot of How To’s from my peers, but I would love more How To Nots.

I really want to see Microsoft take staggered feature adoption seriously. Admin toggles are not scalable. It’s not an easy task, but I think we need something similar to roles or RBAC. Something like Power BI workspace roles, but much, much bigger. The number of Fabric items you can create is 5x the number of Power BI items and growing every day. There needs to be a better middle ground than “turn it all off” or “Wild West”.

One suggestion made by the original LinkedIn author was a paid addon for Power BI pro that adds Power BI Copilot. I think we absolutely do not need that right now. Copilot is expensive in Fabric ($0.32 -$2.90 per day by my math) and still could use some work. It needs more time to bake as LLM prices plummet. If we are bringing Fabric features to a shared capacity model, let’s get Fabric Per User and let’s do it right. Not a rushed job because of some AI hype.

Also, I don’t get why people are expecting a copilot addon or FPU license already. It was 4 years from Power BI Premium (2017) to Premium Per User (2021). It was 3 years from Paginated reports in Premium (2019) until we got Paginated reports in Pro (2022). Fabric has been out for less than 2 years and it is having a lot of growing pains. Perhaps we can be more patient?

How I hope to help

People are reasonably frustrated and feeling lost. Personally, I’d love to see more content about real, lived experiences and real pain points. But complaining only goes so far. So, with that I’m excited to announce the Figuring Out Fabric podcast coming out next week.

You and I can be lost together every week, together. I’ll ask real Fabric users some real questions about Fabric, and we’ll discuss the whole product, warts and all. If you are mad about Fabric, be mad with me. If you are excited about Fabric, be excited with me.