Introduction
Power Query has been the backbone of data transformation in Power BI since the platform's earliest days. It is the layer where raw data gets shaped, cleaned, and prepared before it reaches a report or a dashboard and for most Power BI environments, the quality of that transformation layer determines the quality of everything that sits above it.
In 2026, that layer is changing significantly.
Microsoft has been shipping a series of enhancements across Power Query and Dataflow Gen 2 in Microsoft Fabric that affect performance, usability, and how teams manage reusable transformation logic. For organisations running Power BI at any meaningful scale, whether across a single department or an enterprise data environment, these changes are worth understanding before the gaps they address start showing up in your reporting infrastructure.
This article covers the four most significant developments: the new modern UI coming to Power BI Desktop, the Dataflow Gen 2 performance story, the revised compute unit pricing model, and My Queries: a personal query library that changes how transformation logic gets reused across projects.
What Power Query Is and Why the Transformation Layer Matters
Before getting into what has changed, it is worth establishing why it matters.
Power Query is Microsoft's data preparation and transformation tool which is available across Power BI Desktop, Excel, Microsoft Fabric, Power Platform, and several other environments. At its core it does three things: it connects to data sources, it transforms that data using a language called M Code, and it loads the result to a destination whether that is a Power BI data model, a Fabric lakehouse, or an Excel table.
Dataflows extend this concept to the cloud. Where Power Query in Desktop runs locally on a machine, a Dataflow runs the same transformation logic as a cloud-based service by making it shareable, schedulable, and independent of any individual user's machine.
For organisations building Power BI environments that need to refresh reliably, scale across multiple reports, and serve data to multiple consumers, Dataflows are the more robust architectural choice. The recent enhancements to Dataflow Gen 2 in Microsoft Fabric make that case significantly stronger than it was twelve months ago.
A Unified Modern UI Is Coming to Power BI Desktop
One of the most requested changes in the Power BI community for years has been a consistent Power Query experience across Desktop and the web. The online experience available in Fabric, the Power BI Service, and Power Platform has had a significantly more modern interface than what Power Query in Desktop provided. That inconsistency created friction for teams moving between environments.
Microsoft has started closing that gap. The new modern Get Data experience in Power BI Desktop, released as a preview feature in May 2026, is the first phase of a broader convergence project. Teams can enable it now through Preview Features in Desktop settings.
What the new modern UI brings in this first phase includes an improved connector selection flow, OneLake catalog integration that surfaces available data sources within Fabric, a cleaner navigator experience, and better visual indicators for query folding where the mechanism that determines whether data transformation is being pushed down to the source system or processed locally.
This last point matters more than it might appear. Query folding is one of the most significant performance variables in a Power BI data model. Transformations that fold to the source run at the source's speed, with the source's compute resources. Transformations that do not fold run in Power Query's engine on the local machine or cloud service. Surfaces that make query folding status visible earlier in the development process help teams make better architectural decisions before those decisions become expensive to reverse.
The full modern Power Query editor, not just the Get Data experience is on Microsoft's public roadmap for later in 2026. When it arrives, it will complete the convergence of Desktop and online Power Query experiences, giving teams a consistent interface regardless of which environment they are working in.
For organisations managing Power BI environments across multiple tools and environments, this consistency reduces training overhead and makes it easier to maintain governance standards across the full data preparation layer.
Dataflow Gen 2 Performance: What Has Actually Changed
The performance story behind Dataflow Gen 2 in Microsoft Fabric is one of the most substantive improvements in the Power BI ecosystem in recent memory. Microsoft has published documentation supporting performance improvements of up to 21 times faster execution compared to earlier Dataflow experiences and the mechanisms behind that improvement are worth understanding separately.
Partition Compute
Partition Compute is one of the most impactful performance features in Dataflow Gen 2. It works by identifying tasks within a single M document, a single Power Query mashup, that can be executed in parallel rather than sequentially. Where a standard execution would process transformations one after another, Partition Compute runs independent tasks simultaneously, using available compute resources more efficiently.
The degree of parallelisation is configurable. In the Dataflow options dialogue there is a Scale section with a Concurrency slider, this controls how many tasks can run in parallel at once. The default is tuned for common capacity sizes, but teams running larger or more complex Dataflows can adjust this to match their Fabric capacity and workload requirements.
The Modern Evaluator Engine
The Modern Evaluator is a separate improvement that operates closer to the Power Query engine itself. It represents a migration to a newer .NET runtime for executing M Code in the cloud and it delivers approximately 15 to 20 percent faster execution compared to the previous runtime.
This improvement is available in Dataflow Gen 2 in Fabric and is enabled by default for Dataflows created after March 2026. For Dataflows created before that date, it requires a manual one-click activation in the Options dialogue under Data Flow, Scale, Modern Evaluator Engine. For any team running Dataflows that predate March 2026, enabling this setting is one of the simplest performance improvements available with no configuration required beyond the toggle itself.
The Revised Compute Unit Pricing Model
One of the historical objections to Dataflow Gen 2 adoption was the compute unit consumption model which applied a flat multiplier for the entire duration of a Dataflow run. Microsoft revised this model in late 2025. The current structure applies to the standard multiplier for the first ten minutes of a Dataflow run, then drops to a significantly lower rate for the remainder of the execution time.
Combined with the performance improvements that reduce total execution time, this creates a compounding benefit, the Dataflow finishes faster and the cost-per-run is lower under the new structure than under the old one. For teams that evaluated Dataflow Gen 2 twelve months ago and found the compute economics unattractive, the current model is worth revisiting.
My Queries: A Personal Query Library in Fabric
My Queries is a feature in public preview that solves a problem most Power Query practitioners have encountered and worked around in various ways for years.
In any Power BI or data team, certain queries get written repeatedly. A date table generation query. A custom function for parsing a specific data format. A helper query that cleans a particular field type. These reusable pieces of transformation logic tend to live in text files, GitHub repositories, personal notebooks, or are simply rewritten from memory each time they are needed.
My Queries provides a personal library within Microsoft Fabric where these queries can be saved and retrieved directly within the Dataflow Gen 2 authoring experience. Rather than locating an old file, copying the M Code, and pasting it into a new Dataflow, a saved query can be imported directly from the library with a few clicks.
Several aspects of how this feature works are worth understanding clearly.
The library is personal; it saves the individual user's personal workspace in Fabric rather than to a shared team workspace. This makes it immediately useful for individual practitioners and removes the need for external workarounds, while also meaning that team-level query sharing requires a different approach for now.
The library supports any Power Query object not just tables. Functions, lists, custom logic patterns, helper scripts where any valid M Code object can be saved. This is particularly valuable for teams that have built libraries of custom functions over time and have previously had no structured way to manage and reuse them within Fabric.
The feature is available now as public preview for Fabric capacity workspaces, with editing, updating, and deletion capabilities evolving as the feature matures.
For Power BI teams that have been managing reusable query logic informally, My Queries provides the first structured, platform-native approach to query reuse within Fabric's Dataflow environment.
The SharePoint List Picker: A Small Change That Removes Real Friction
One smaller enhancement worth noting is the new SharePoint List Picker in the SharePoint connector. Previously, connecting to a SharePoint site required knowing and entering the site URL, a step that created unnecessary friction for users who knew which site they needed but not its exact URL structure.
The new experience surfaces a list of SharePoint sites the user has access to, allowing selection by name rather than URL. It is a minor change in technical terms but a meaningful one in practice, particularly for teams connecting Power BI reports to SharePoint-hosted data sources, where the URL requirement created a consistent point of user error and support overhead.
What This Means for Organisations Running Power BI
The cumulative effect of these changes is a Power BI data transformation layer that is faster, more consistent across environments, and better equipped for reuse at team scale.
For organisations where Power BI is connected to systems like NetSuite through NetSuite and Power BI integration services, the Dataflow Gen 2 performance improvements directly affect how quickly financial and operational data can be refreshed and made available to report consumers. Faster Dataflows mean more current data in the hands of the people making decisions.
For finance transformation initiatives where Power BI is the reporting layer over restructured financial data, the improved query folding visibility and the unified UI reduce the risk of transformation decisions that look correct in development but perform poorly in production.
And for data and technology teams managing multiple Power BI environments, the My Queries library and the unified UI reduce the inconsistency and duplication that accumulates when transformation logic is managed informally across individuals rather than through a structured platform capability.
Three things worth doing now if your organisation is running Power BI with Dataflow Gen 2.
Enable the Modern Evaluator Engine on any Dataflow created before March 2026. It is a one-click improvement worth ten to twenty percent faster execution and requires no configuration changes to the transformation logic itself.
Enable the modern Get Data experience in Power BI Desktop Preview Features. It is opt-in at this stage and does not affect existing queries or models but it gives teams access to the new UI and surfaces the query folding indicators that make better architectural decisions easier.
Evaluate My Queries for any reusable transformation logic your team currently manages outside the platform. The feature is in preview and will evolve but the value of moving query libraries from external files into a platform-native location compounds quickly for teams that work across multiple Dataflows and multiple projects.
For teams looking at the broader question of how to build and govern their Power BI environment, including how to connect AI capabilities to your existing Power BI infrastructure, our blog on connecting Power BI to Claude AI using the MCP Server covers how the MCP layer enables AI-assisted DAX generation and model interrogation directly inside your Power BI file.
Closing
Power Query and Dataflow Gen 2 are not features that get talked about as much as dashboards, DAX, or AI integration. They sit below the visible layer in the transformation and preparation work that determines whether the data reaching a report is trustworthy, current, and scalable.
The 2026 enhancements to this layer, the modern UI convergence, the Partition Compute and Modern Evaluator performance improvements, the revised pricing model, and My Queries represent meaningful progress on problems that have existed in the platform for years.
For organisations where the quality of business decisions depends on the quality and currency of Power BI data, understanding and adopting these changes is not optional. It is the kind of platform maintenance that keeps a Power BI environment performing at the level the business expects from it.
