Microsoft launched Microsoft Fabric, a unified analytics platform that combines technologies like Data Factory, Synapse and PowerBI. Microsoft's pitch: Convince enterprises to go with one integrated AI-powered analytics platform instead of integrating multiple vendors and services.

The news, launched at Microsoft's Build conference, is the headliner, but there was a bevy of other Copilot announcements. Microsoft also announced Copilot in Power BI, Power BI Direct Lake, a new storage mode, and Power BI Desktop Developer Mode. Speaking during the Build 2023 keynote, Microsoft CEO Satya Nadella said "platform shifts are in the air." Speaking about generative AI, Nadella said "every piece of the stack has been impacted" and there were 50 announcements related to ChatGPT on deck. 

Regarding Microsoft Fabric, Nadella said:

"This is a product we've been working hard on over multiple years. This is the biggest data product launch since the launch of SQL Server. It unifies the business model across all analytics workloads. This unification will fuel the next generation of AI applications." 

Constellation Research analyst Doug Henschen put Microsoft Fabric in context:

"We’ve seen a bunch of fabric-type offerings and customers seem to be keen on the idea of having one platform that provides access to all data – even if, in actuality, it’s about distributed data access layer on top of multiple repositories and capabilities, as is clearly the case with Microsoft Fabric. Other examples recently in the news include IBM Data Fabric and its acquisition of Ahana and SAP Datasphere. Other incumbents include Dremio, Starburst, and various “lakehouse” offerings, though Microsoft is combining a huge breadth of workload types and is promising generative AI interfaces on top of them all."

Microsoft Data Fabric also lands as multiple vendors are aiming to be the business process automation platform of choice.

The bet here is that Microsoft's pricing strategy can make Fabric compelling and more economical for CXOs. Customers can buy one pool of compute to power Fabric workloads. That pricing strategy could be compelling. Henschen said:

"One of the most compelling aspects of this announcement, in my view, was the prospect of buying a single pool of credits that can be used across all workload types: data engineering, data integration, data science, data warehousing, BI and analytics. That’s a unique offering and gives a sense of a single platform, even if it’s, technically, a unified layer on top of a bunch of existing offerings. This is classic suite versus best-of-breed marketing. The complication is that these budgets are owned by different groups today, and they’d have to get together to make decisions about how many credits each group needs and so on."

Ferguson, T-Mobile and AON were named customers looking to consolidate their analytics footprints on Fabric. Informatica is one of the first design partners for Fabric and the company said customers can enroll in a private preview starting in June 2023. 

Related researchAnalytics and Business Intelligence Evolve for Cloud, Embedding, and Generative AI | Analytics and Business Intelligence Trends in Cloud, Embedding, and Generative AI | Constellation ShortList™ Multicloud Analytics and Business Intelligence Platforms

Key points about Microsoft Fabric include:

  • Fabric will have a single unified experience and architecture via a SaaS delivery model. It will also integrate with Microsoft 365 applications.
  • Each team in the analytics process including data engineers, data warehousing pros, data scientists, data analysts and business users will have a role-specific experience.
  • Fabric is data-lake friendly and open. Microsoft Fabric includes OneLake, a multi-cloud data lake that is built-in. The model for OneLake rhymes with the OneDrive approach. OneLake is built on Azure Data Lake Storage Gen2 and fully compatible. OneLake also has shortcuts to data lake storage from Azure, AWS S3 and Google Storage coming in the near future.
  • OneLake supports structured data of any format and unstructured data. OneLake will also be discoverable and accessible in Microsoft 365.
  • Microsoft Fabric will include Azure OpenAI Service at multiple layers. Generative AI and Copilot in Microsoft Fabric will be available and work with various models. Copilot in Microsoft Fabric will be coming soon.

Henschen noted that natural language (NL) can be the interface of Fabric. he said:

"I think we can expect to see Co-Pilot interfaces proving NL interaction based on the Azure OpenAI service for every type of user. Data engineers will have options to use NL to generate code and drive Spark Workloads. Data integrators and analysts will use NL to generate SQL data transformations and SQL queries. Analysts and business users will ask questions to generate visualizations and dashboards. The workloads and platform are basically the same, but the generative AI lets you, it is promised, work more efficiently and productively without all the coding and drudgery that was previously required."

The workloads

Microsoft said Fabric comes with seven workloads in preview.

  • Data Factory has more than 150 connectors to cloud and on-prem data sources and the ability to transform data and orchestrate pipelines.
  • Synapse Data Engineering enables authoring in Spark, instant start with live pools, collaboration tools.
  • Synapse Data Science provides workflows for building models, collaborating, training and deploying them.
  • Synapse Data Warehousing provides a converged lake house and data warehouse experience SQL performance on open data formats.
  • Synapse Real-Time Analytics analyzes streaming data from IoT and edge devices, telemetry, logs with low latency.
  • Power BI in Microsoft Fabric provides visualization and analytics. Power BI in Fabric is also integrated into Microsoft 365 apps.
  • Data Activator monitors data and can trigger notifications and actions.

Finally, there's the caveat: Everything announced is still in preview. Henschen said:

"I think we’ll see a lot of wait-and-see reactions from customers and a lot of competitive responses. Changing data platforms is akin to turning a super tanker: It’s not something that happens quickly, and I would not expect a lot of market movement overnight."