From Ledgers to Intelligence Part 17: The Unified Intelligence Platform When BI, AI, and Data Engineering Converge

Digital Transformation | June 2026

The Modern Data Stack assembled a best-of-breed architecture from specialised, interoperable tools: Fivetran for ingestion, Snowflake for storage, dbt for transformation, Looker for semantics and consumption. For organisations that could staff and integrate these tools, it delivered remarkable analytical capability. For organisations that lacked the engineering bandwidth to assemble and operate a multi-vendor stack or for enterprise technology leaders tired of managing the integration seams between them a different proposition was emerging.

By 2024, the major cloud platform providers were converging on a unified vision: a single platform that collapsed the modern data stack’s specialised layers into a coherent, integrated product. Data engineering, data warehousing, analytics, machine learning, and AI all in one surface, from one vendor, on one governance model. The era of the unified intelligence platform had begun.

Microsoft Fabric: OneLake as the Unifying Layer

Microsoft Fabric, announced at Microsoft Build 2023 and generally available from November 2023, is Microsoft’s most ambitious data platform bet since the original Azure launch. Its central concept is OneLake a single, tenant-wide data lake built on Azure Data Lake Storage Gen2, accessible to all Fabric workloads simultaneously. Data engineering pipelines write to OneLake; Fabric’s data warehouse queries from OneLake; Power BI reports from OneLake; Fabric’s ML workloads train from OneLake. There is one copy of the data, in one place, with one security and governance model (Microsoft Purview).

Fabric’s integration extends to the development surface. A data engineer working in a Synapse Spark notebook, a data analyst building a Power BI report, and a data scientist training a model in Fabric’s ML workspace are all working within a single product, with shared compute billing, shared workspace management, and shared governance. Copilot Microsoft’s LLM-powered AI assistant is embedded throughout, generating code, explaining results, and suggesting transformations.

For organisations already deep in the Microsoft ecosystem Azure, SQL Server, Power BI, Teams, Office 365 Fabric’s integration advantages are substantial. The migration path from Azure Synapse Analytics and Power BI Premium is well-defined. The trade-off is the usual Microsoft trade-off: deep integration in exchange for reduced portability.

Databricks: The Intelligence Platform

Databricks, founded by the creators of Apache Spark and Delta Lake, rebranded its offering in 2023 as the “Databricks Intelligence Platform” a positioning shift from “big data and ML platform” to “unified analytics and AI.” The platform assembles Delta Lake (storage), Apache Spark (compute), MLflow (ML lifecycle), Unity Catalog (governance), and Databricks SQL (interactive analytics) into a coherent product, with Databricks Genie an LLM-powered natural language data interface as the AI layer.

Databricks’ strategic differentiation from Microsoft Fabric is openness: Delta Lake, MLflow, and Unity Catalog are open-source projects governed by the Linux Foundation. Data stored in Delta Lake format in a customer’s own cloud storage account is readable by any compatible engine. Databricks is the governance and compute layer; the data remains portable. This positioning appeals to organisations concerned about vendor lock-in and to engineering teams that value the ability to use multiple tools against the same data.

Google: BigQuery + Looker + Vertex AI

Google’s unified data and AI platform assembles three products that were developed or acquired separately: BigQuery (the cloud data warehouse, since 2010), Looker (acquired 2019 for $2.6 billion), and Vertex AI (Google Cloud’s managed ML platform, 2021). The integration between these products has deepened steadily BigQuery ML allows ML model training in SQL, Vertex AI Feature Store uses BigQuery as its offline store, and Looker semantic models can be queried by Vertex AI agents.

Google’s differentiation is technical depth: BigQuery’s serverless architecture and Gemini-powered capabilities (BigQuery AI queries, natural language to SQL via Duet AI) represent genuine engineering innovation. Looker’s semantic layer, though sometimes criticised for sluggish product development post-acquisition, remains architecturally sophisticated. The weakness is cohesion: the three products still feel like three products rather than one.

Composable vs Integrated: The Strategic Choice

The emergence of unified intelligence platforms does not render the composable modern data stack obsolete it creates a strategic choice that every data team must make explicitly. The composable approach (best-of-breed tools assembled by engineering) offers maximum flexibility, tool-level optimisation, and the ability to replace any component independently. The integrated approach offers reduced operational complexity, tighter product cohesion, unified billing and support, and lower engineering overhead.

The right choice depends on team size and capability, existing vendor relationships, data volume and complexity, and the organisation’s tolerance for vendor lock-in versus operational simplicity. A 50-person technology company with strong data engineering capability may extract more value from a composable stack. A 5,000-person enterprise with a small data team and a Microsoft-heavy technology estate may be better served by Fabric. Neither is universally correct.

---

References

1. Microsoft Corporation (2023). Introducing Microsoft Fabric: Data and analytics for the era of AI. Microsoft Build 2023.
2. Databricks (2023). The Databricks Intelligence Platform: Simplify your data and AI. Databricks Summit 2023.
3. Google Cloud (2023). BigQuery and Vertex AI: Building an enterprise data and AI platform. Google Cloud Next 2023.
4. Gartner Research (2024). Magic Quadrant for Analytics and Business Intelligence Platforms.
5. Snowflake Inc. (2024). Snowflake Cortex AI: Enterprise AI built on your data. Snowflake Summit 2024.
6. Stancil, B. (2023). The Composable CDP and the End of the MDS. Benn Stancil Substack.
7. Databricks (2023). Unity Catalog: Open source governance for the data lakehouse. Linux Foundation Announcement.