I’ve been speaking to loads of Energy BI builders within the final 12 months who’re really anxious about Material. Reality to be mentioned, the messaging from Microsoft has been hype-heavy, and the group has been shifting quick. And each time you open LinkedIn or Reddit, someone is saying a brand new Material characteristic you’ve by no means heard of, with a screenshot of a Lakehouse you don’t perceive, speaking about MLVs and Direct Lake on OneLake and Material IQ as if everybody already is aware of what these issues are.
I get it. It appears like the bottom is shifting below your toes.
So, the very first thing I’m telling everybody (and all of you who’ve ever attended my trainings/dwell periods already heard this from me): your Energy BI abilities are intact. Your experiences nonetheless work. Your DAX issues greater than ever. And you do not want to be taught Spark this weekend.
What you DO want is a map. One which tells you what truly modified for YOU as a Energy BI developer, what you may safely ignore for now, and the place to start out while you’re able to take the subsequent step. This text goals to be that map.
What didn’t change (begin right here and breathe)
Earlier than we speak about something new, let me affirm what’s nonetheless precisely the identical. As a result of in all of the noise about Material, this half typically will get ignored.
Energy BI’s authoring expertise remains to be the identical. Whether or not you construct fashions and experiences in Energy BI Desktop or within the browser utilizing internet modeling (which is now primarily at parity with Desktop), the workflow hasn’t modified. You continue to load knowledge, construct a mannequin, write DAX, design a report, and publish. Identical abilities, similar patterns.
DAX remains to be “the language” for semantic fashions, and it’s not going wherever. If something, it’s MORE vital now — Copilot generates DAX, Direct Lake queries run on DAX, and each new AI characteristic in Material is determined by well-written measures. The higher your DAX, the higher all the pieces downstream works.
Energy Question and the M language nonetheless deal with knowledge transformation throughout Desktop, internet modeling, and Dataflows. And, though the outdated Dataflows Gen1 shall be formally deprecated, the Gen2 model is right here to remain, and it depends on Energy Question/M abilities. Your current data of Energy Question patterns — merging, appending, conditional columns, parameters — all nonetheless applies.
Semantic fashions are nonetheless the identical factor. Identical VertiPaq columnar storage, similar relationships, similar calculated columns, similar measures, similar Greatest Apply Analyzer guidelines…
The Energy BI service is now a part of the Material portal, however when you log in tomorrow morning, you’ll discover the identical workspaces, the identical experiences, the identical scheduled refreshes, the identical Apps expertise. Even the URL didn’t change.
Import mode and DirectQuery nonetheless work precisely as they did earlier than. You don’t HAVE to make use of Direct Lake. In case your experiences run high quality on Import mode with a every day refresh, that’s a wonderfully legitimate structure in 2026.
Row-level safety, deployment pipelines, gateways, Apps — all nonetheless there. Nothing was eliminated.
OK, so your basis is stable. However some issues DID change, and some of them truly matter to you. Let me stroll you thru those which are value your time.
What modified and truly issues to you
I’ve recognized 5 issues. In case you perceive these 5 modifications, you perceive 90% of what’s completely different about being a Energy BI developer in Material.
1. Your license modified (and it’s high quality)
In January 2025, Microsoft retired the Energy BI Premium P SKUs. In case your group was operating on Premium — P1, P2, P3 — your surroundings has already been moved to Material capability (F SKUs). P1 is roughly equal to F64. P2 is roughly F128, and so forth.
The important thing factor to recollect is that nothing broke throughout that transition. Stories work, semantic fashions work, paginated experiences work, scheduled refreshes work… Your current funding is preserved.
What you GAINED was entry to the complete Material platform below the identical license — Lakehouses, Warehouses, pipelines, Actual-Time Intelligence, all of it. You don’t have to make use of them, however they’re sitting there, included, prepared while you need them.
One factor to remember: under F64, report customers nonetheless want Energy BI Professional licenses to view content material. At F64 and above, viewer entry is included within the capability license, which is why F64 is the business threshold most organizations goal if they’ve greater than 300 report customers.
2. OneLake exists, and your knowledge lives there now
OneLake is the unified storage layer for all the pieces in Material. Consider it as “OneDrive for knowledge.” One lake. One location. One copy.
Once you create a Lakehouse, a Warehouse in Material, the info is saved in OneLake as Delta format — the open desk format that’s change into the trade normal for analytics. Your semantic fashions, Lakehouse tables, Warehouse tables, all of them dwell in the identical place beneath.
Why does this matter to you as a Energy BI developer? Due to the subsequent change…
3. Direct Lake is a brand new storage mode possibility
You already know two storage modes: Import (load knowledge into the mannequin, refresh on a schedule) and DirectQuery (question the supply dwell each time). Every has trade-offs you’ve in all probability discovered the onerous manner: Import is quick however stale, DirectQuery is recent however sluggish.
Direct Lake is a 3rd possibility. It reads Delta tables immediately from OneLake into reminiscence — no scheduled refresh, no dwell question overhead. Quick queries AND recent knowledge. Direct Lake is the closest factor to “you may have each” that Energy BI has ever provided.
The important thing factor to remember is that the expertise of constructing experiences doesn’t change. You continue to creator DAX the identical manner, you continue to design visuals the identical manner. The mechanics beneath modified (your mannequin is studying immediately from Delta in OneLake as an alternative of holding its personal copy), however as a developer, the workflow appears the identical.
When do you have to care about Direct Lake? In case your semantic fashions are giant (hundreds of thousands of rows, multi-GB fashions) and refresh instances are painful, Direct Lake is value investigating. In case your fashions are small and refresh takes 30 seconds, Import mode remains to be completely high quality. No urgency.
I wrote concerning the two flavors of Direct Lake — Direct Lake on SQL and Direct Lake on OneLake, so you might need to learn that one as properly.
4. Copilot is in every single place
Copilot in Energy BI has expanded considerably within the final 12 months. You’ll discover it in Desktop (generates report pages, generates DAX, suggests measures, explains formulation), within the service (report-level chat, creates visuals from pure language), and on cell (as of April 2026, conversational report exploration in your telephone or pill).
I constructed an total Pluralsight course on Copilot in Energy BI, so right here’s my trustworthy take: it’s helpful, not magical. Copilot is basically good at producing starter DAX measures, summarising what’s on a report web page, and serving to you ask exploratory questions of a mannequin. It’s NOT that good at writing complicated measures from scratch, understanding nuanced enterprise logic, or working with poorly-designed semantic fashions.
The sample I’ve seen is constant: the higher your semantic mannequin, the higher Copilot performs. Good desk names, good column names, correctly outlined measures, clear relationships… If these are in place, Copilot turns into an actual productiveness device. In the event that they’re not, Copilot generates plausible-looking nonsense.
What you may safely ignore (sure, I actually imply it)
Now, the part that no person else will write for you:) The Material ecosystem is big: Spark, KQL, Actual-Time Intelligence, Knowledge Manufacturing facility, Knowledge Activator, MLVs, Material IQ, Knowledge Brokers… And the social media-driven stress to be taught ALL of it RIGHT NOW is misplaced.
You possibly can safely ignore most of it. I’ll attempt to share the breakdown:
Spark notebooks and PySpark. Except your position is increasing into knowledge engineering, you don’t want these. Your Energy Question abilities are nonetheless the proper device for Energy BI knowledge prep. If a knowledge engineer in your group builds a Spark pocket book that lands clear Delta tables in OneLake — nice. You devour them. You don’t want to jot down them.
KQL databases and Actual-Time Intelligence. Solely related when you’re working with streaming knowledge, IoT telemetry, or high-volume occasion logs. Most PBI builders aren’t. In case your knowledge lives in a database and refreshes every day, Actual-Time Intelligence shouldn’t be your drawback.
Knowledge Activator. Occasion-driven automation. Actually, it’s very fascinating, however not core to report improvement. File this below “discover when curiosity pulls you there.”
Material IQ and Knowledge Brokers. The brand new AI/pure language layer that lets customers ask questions of their knowledge in plain English. Price maintaining a tally of, however you don’t have to architect round it but.
Materialized Lake Views (MLVs). Good for knowledge engineers constructing medallion architectures. Not one thing Energy BI builders have to create, although you’ll fortunately devour the silver and gold MLVs that knowledge engineers construct for you. In case you’re curious, my MLV deep dive explains them intimately. However you don’t have to learn it to maintain doing all of your job properly.
The important thing message: be taught these when your position calls for it or when your curiosity genuinely pulls you towards them. Not as a result of the algorithm advised you to be scared.
The “begin right here” path
OK, you’re able to take the primary concrete step. Right here’s the sequence I’d suggest — the one I give to Energy BI builders on my group once they ask the place to start.
Step 1: Perceive your capability. Discover out whether or not your group is on Material capability (an F SKU), Energy BI Professional, or Premium Per Consumer. This single reality determines what’s out there to you.
Step 2: Discover OneLake and a Lakehouse. Create a Lakehouse in a workspace. Add a CSV file. Watch it change into a Delta desk. Question it from the SQL analytics endpoint. Join Energy BI to it — from Desktop or immediately within the browser, your selection. This single hands-on train demystifies 80% of the Material terminology you’ve been seeing. The most effective factor? It takes a day.
Step 3: Strive Direct Lake on a non-production dataset. Create a semantic mannequin out of your Lakehouse utilizing Direct Lake storage mode. Construct a small report. Discover that there’s no refresh button — you simply publish, and the info is at all times present. Examine question efficiency to your current Import mannequin. Get a really feel for the way Direct Lake behaves.
Step 4: Let Copilot provide help to with one thing actual, however confirm. Open Copilot — it’s out there wherever you creator Energy BI content material. Ask it to generate a DAX measure for one thing particular to your mannequin. Test its work. Iterate. Then ask Copilot to summarise a report web page and see if the abstract is correct. Construct your personal sense of the place Copilot is reliable and the place it isn’t.
Every step takes a day. Not a month or quarter.
The profession query no person needs to ask
And, to wrap this up, let me deal with the elephant within the room: “Is my Energy BI profession protected?”
Brief reply: sure. However the lengthy reply is extra fascinating and extra nuanced. I already wrote an article about the way forward for Energy BI from the speedy rise of AI-generated dashboards. Right here, I need to concentrate on the Energy BI future, trying by way of the Material lenses.
Energy BI builders are NOT being changed by Material builders. The roles are completely different. Energy BI improvement — semantic modeling, DAX, report design, business-facing analytics, is a definite and worthwhile skillset that the platform essentially is determined by.
What IS altering is that the very best Energy BI builders in 2026 would be the ones who perceive the place their knowledge comes from (OneLake, Lakehouses, Warehouses) and the way it will get there. You don’t have to BUILD the pipeline. However understanding the structure round your experiences makes you considerably simpler and considerably extra worthwhile in conversations with knowledge engineers, architects, and stakeholders.
If you’re searching for certifications as a technique to show your abilities, there are a couple of certifications to maintain in your radar. The PL-300 (Energy BI Knowledge Analyst) remains to be the core PBI certification and is being stored present — it’s nonetheless the inspiration. The pure subsequent step for Energy BI builders shifting into Material is the DP-600 (Material Analytics Engineer Affiliate). DP-600 bridges semantic modeling, DAX, and the broader Material platform from a BI perspective. If colleagues in your group are heading into knowledge engineering, DP-700 (Material Knowledge Engineer Affiliate) is their path — completely different position, completely different examination. For Energy BI builders, PL-300 adopted by DP-600 is the mix I’d suggest.
One other level that’s value saying out loud: your DAX and modeling abilities are MORE worthwhile in Material, not much less. Direct Lake efficiency is determined by well-designed Delta tables and clear semantic fashions. Copilot performs higher with clear fashions. Knowledge Brokers generate higher solutions from clear fashions. Material IQ surfaces higher insights from clear fashions. EVERYTHING downstream is determined by the standard of the semantic layer, and the semantic layer is YOUR area.
The most effective profession transfer you can also make proper now isn’t studying Spark. It’s getting even higher at semantic modeling and DAX. Grasp Energy BI’s modeling fundamentals, get fluent in superior DAX patterns, learn to design fashions for Direct Lake efficiency, after which, while you’re prepared, add Material context round it.
Wrapping up
Material isn’t a alternative for Energy BI. It’s the platform Energy BI now lives inside. Your abilities are intact. Your experiences work. Your DAX issues greater than ever. And also you don’t have to be taught the whole Material stack to be wonderful at your job.
Begin with the 4 steps. Take them at your personal tempo, ignore the components that don’t apply to your position but, and if you wish to go deeper, I’ve been writing about Material structure, Direct Lake, materialized lake views, and the Warehouse vs. Lakehouse choice — all from a working architect’s perspective.
You haven’t missed the bus. You’re on it. The vacation spot is just a bit additional than you thought.
Thanks for studying!
Because of Claude for creating these nice-looking illustrations.
