Ten years. I keep counting it on my fingers like it might come out different the next time. It does not.
In early 2017, Microsoft awarded me the Data Platform MVP for the first time. I had been speaking at community events for about four years at that point, starting with a shaky user group presentation in Copenhagen and gradually building up to SQLSaturdays across Europe. When the notification mail arrived I read it three times. Then I called my wife. Then I read it again.
What I did not anticipate was how much the platform itself would transform in the decade ahead, or how completely my own technical identity would change alongside it. This is a reflection on both.
2016: On-premises and proud of it
When the MVP journey began, my world was SQL Server. Specifically, Analysis Services Multidimensional. I wrote MDX for a living, tuned partition strategies, scripted deployments in PowerShell, and built dimension security models with AMO. The toolchain was mature, the community was tight, and the assumption was that your servers lived in a rack somewhere you could point to.
Azure existed, of course. It was about six years old. But for most of us in the BI world it still felt like someone else’s problem. Azure SQL Database was limited. Azure Analysis Services was in preview. The idea that a serious analytical workload could run entirely in the cloud was something you heard at conference keynotes and politely ignored on Monday morning.
I remember attending a PASS Summit session around that time where someone demoed Azure Data Lake by showing a slide deck of screenshots. With errors in them. That was the state of cloud analytics in our corner of the Microsoft ecosystem. Promising on paper. Not quite there in practice.
2017 to 2020: The cloud came whether we were ready or not
Each year I renewed the MVP award, Azure had added another service that pulled more of my on-premises workload toward the cloud. Azure Data Factory replaced chunks of what SSIS used to do. Azure Synapse started absorbing the data warehouse conversation. Managed identities appeared, and with them a new category of debugging I had never encountered before: authentication that worked in test and mysteriously failed in production with no clear error message.
The learning curve was not optional. I recertified Azure exams not for the credential itself but because the platform had changed enough that I needed to confirm I still understood what I was deploying. Every renewal cycle felt like a different job.
During this same stretch, Power BI went from “promising” to “the center of gravity for everything Microsoft BI.” It shipped in mid-2015, and within two years the conversation at every community event had shifted from “should we look at Power BI?” to “how do we govern hundreds of workspaces?” I took over running the Danish Power BI User Group and watched it grow past 2,000 members. The problems I solved daily moved from MDX tuning to DAX optimization, from partition scripting to deployment pipelines and semantic model management. Different skill set. Same instinct: figure it out, test it, share what happened.
2021 to 2023: The convergence
Then the pieces started merging. Azure Synapse tried to be the unified analytics service. Power BI Premium brought dataset hosting into the cloud. Microsoft kept drawing a tighter circle around what had been separate products, pushing toward something integrated.
And then Fabric arrived.
Announced at Build in May 2023, generally available by November that year. Within months it had absorbed the Power BI service, introduced Lakehouses, OneLake as a unified storage layer, Data Pipelines, and Notebooks as a first-class authoring experience. The pitch was simple: one platform for data engineering, data science, real-time analytics, and business intelligence. No more stitching five Azure services together and hoping the authentication tokens lined up.
2024 to now: A different job
The speed of Fabric’s evolution catches me off guard sometimes. I spent years becoming deeply skilled in SSAS Multidimensional. That knowledge still informs how I think about models, but it is no longer the center of my work. The architecture underneath shifted fundamentally. Lakehouses replaced cubes. Notebooks replaced XMLA scripts. Semantic models became the new version of what we used to call datasets, which themselves replaced what we used to call cubes.
Each rename is not just marketing. Each one reflects a genuine change in how the technology works and what you can build with it.
Today I write Python in Fabric Notebooks to snapshot and diff semantic model metadata. Materialized Lake Views went GA. Fabric IQ brought SQL-familiar tooling to the Lakehouse. I have found myself using AI to reverse-engineer Power BI models and building open-source tooling on top of the Tabular Object Model through semantic-link-labs. The problems I solve in 2026 look nothing like the problems I solved in 2016.
What stayed the same
The community.
That is the thread that runs through all ten years without interruption. The tools changed completely. The community changed shape too, but it did not break. PASS dissolved. SQLSaturday became Data Saturday. The #sqlfamily hashtag gave way to broader data community networks. But the people kept showing up.
The person who initially nominated me for the MVP award: Mark Broadbent (l|b|m|x). The organizer who let me both speak and help run events for years. The group of Danes who went to dinner together after Summit sessions in Seattle. The local organizers in Dublin, Gothenburg, Chicago, Prague, Pittsburgh, New York, Krakov, Stockholm, London, Newport, Cambridge, Hanau, Munich, Thorshavn, Utrecht…, who each time made space for someone to come talk about whatever they had just figured out. Every one of those moments is part of this decade.
I still organize Data Saturday Denmark. Some years it goes smoothly. Some years 95 people register and do not show up, and you have to write the uncomfortable post about it, with new rules and transparent reasoning, because that is what organizing actually looks like: not just running the event but protecting its future.
The parallel timelines
If I map my MVP journey against the platform’s, the parallels are hard to miss:
When I got the award, the platform was SQL Server with Azure as a sidecar. My expertise was multidimensional cubes and MDX. The community gathered under the PASS umbrella.
Five years in, the platform was Azure-first with Power BI as the analytical surface. My expertise had shifted to data modeling, deployment pipelines, and cloud administration. The community was regrouping after PASS shut down.
Now, at ten years, the platform is Fabric. My expertise is Lakehouses, semantic models, Notebooks, and AI-assisted development. The community runs on Data Saturdays, LinkedIn, and local user groups that feel smaller but more committed.
Three different jobs descriptions (same place) over one continuous award. The MVP title stayed the same on paper. Everything underneath it turned over at least twice.
What I take from it
I do not have a tidy lesson. I have an observation. The practitioners who stayed relevant through these shifts were not the ones who predicted the roadmap correctly. They were the ones who kept testing things with real data, publishing their mistakes, and showing up to events where they might learn something new. That is what I tried to do. Some years better than others.
Ten years. Same instinct, completely different platform.
