'%3e%3cg%20transform='translate(0%200)'%20clip-path='url(%23a)'%3e%3cpath%20d='M115.449,63.552a36.148,36.148,0,0,1-5.422-.418A43.162,43.162,0,0,1,86.866,51.671L65.554,31.142,49.589,47.105a4.338,4.338,0,1,1-6.136-6.132L59.3,25.118l-6.5-6.266A4.342,4.342,0,1,1,58.82,12.6l6.623,6.377L81.894,2.521a4.356,4.356,0,0,1,6.147,0,4.36,4.36,0,0,1,0,6.144L71.7,25,92.853,45.379a34.433,34.433,0,0,0,18.512,9.167,26.307,26.307,0,0,0,4.047.32,24.116,24.116,0,0,0,10.812-2.548c5.818-2.9,10.991-8.25,15.376-15.906a4.322,4.322,0,0,1,5.92-1.616,4.338,4.338,0,0,1,1.617,5.92c-5.25,9.183-11.661,15.7-19.051,19.377a32.516,32.516,0,0,1-14.636,3.459'%20transform='translate(12.691%200.377)'%20fill='%23fff'%3e%3c/path%3e%3cpath%20d='M155.17,40.416a4.926,4.926,0,1,1,4.926-4.926,4.932,4.932,0,0,1-4.926,4.926'%20transform='translate(45.206%209.196)'%20fill='%23ab00f1'%3e%3c/path%3e%3cpath%20d='M44.06,0a4.346,4.346,0,0,0-4.341,4.341V6.218l-1.3-.952A23.922,23.922,0,0,0,24.2.63a24.2,24.2,0,0,0,0,48.4,23.95,23.95,0,0,0,14.242-4.645l1.2-.882.1,1.49a4.336,4.336,0,0,0,8.663-.3V4.341A4.346,4.346,0,0,0,44.06,0M24.2,40.339A15.514,15.514,0,1,1,39.719,24.825,15.532,15.532,0,0,1,24.2,40.339'%20transform='translate(0%200)'%20fill='%23fff'%3e%3c/path%3e%3cpath%20d='M103.077.481a24.2,24.2,0,1,0,24.2,24.2,24.229,24.229,0,0,0-24.2-24.2m0,39.717a15.516,15.516,0,1,1,15.514-15.516A15.535,15.535,0,0,1,103.077,40.2'%20transform='translate(23.732%200.145)'%20fill='%23fff'%3e%3c/path%3e%3cpath%20d='M164.56,35.221a4.349,4.349,0,0,1-4.344-4.344V23.9a15.209,15.209,0,1,0-30.419,0,4.342,4.342,0,1,1-8.685,0,23.895,23.895,0,0,1,47.789,0v6.982a4.349,4.349,0,0,1-4.342,4.344'%20transform='translate(36.441%200)'%20fill='%23fff'%3e%3c/path%3e%3cpath%20d='M46.655,10.867A4.926,4.926,0,1,1,51.582,5.94a4.932,4.932,0,0,1-4.926,4.926'%20transform='translate(12.556%200.305)'%20fill='%23ab00f1'%3e%3c/path%3e%3c/g%3e%3c/g%3e%3c/svg%3e)
What OneLake is
OneLake is a single, tenant-wide data lake that ships with Microsoft Fabric. Every workload in Fabric (Power BI, Data Factory, Synapse, Real-Time Analytics, Data Science) reads from and writes to the same OneLake. You do not provision it, configure it, or choose where it lives - it is just there the moment your tenant is enabled.
Why Microsoft calls it OneDrive for data
OneDrive solved a real problem: you used to email documents around and end up with five versions, none of which agreed. OneLake aims to do the same for business data. Instead of every system having its own copy of customer, product, or sales data, OneLake holds one version and every tool references it in place. No more copying data from the warehouse to Power BI, from Power BI to Excel, from Excel into a slide. One source, many tools.
Shortcuts: the bit that does the heavy lifting
Shortcuts let OneLake reference data that lives elsewhere (Azure Data Lake, Amazon S3, Google Cloud Storage, Dataverse) without copying it. The data physically stays where it is, but every Fabric workload can query it as if it were inside OneLake. This is the feature that quietly makes most legacy storage problems go away, because you do not have to migrate everything before you can use it.
What changes for your business
Three practical things. First, less duplication: finance, ops, and marketing all report on the same numbers. Second, less ETL: data does not have to be copied between systems before it can be used. Third, faster analytics projects: a new dashboard, a new model, or a new Copilot use case starts from data that already exists and is already trusted, instead of a fresh extract.
What it does not change
OneLake does not magically make messy data tidy. If your source systems disagree on what a customer is, OneLake will faithfully store the disagreement. The platform makes governance easier (one place to define what good looks like) but the work of agreeing on definitions still has to happen.
The open format underneath
OneLake stores everything in the open Delta Lake format. That matters more than it sounds. It means your data is not locked into a Microsoft-only format - other tools that understand Delta (Databricks, Snowflake, open-source Spark) can read OneLake data directly, without exports or copies. If you ever wanted to use a non-Microsoft analytics tool alongside Fabric, you could, and your data would still be in one canonical place. That openness is unusual for a major cloud vendor and is a real factor when comparing Fabric to alternatives that lock data into proprietary formats.
How OneLake handles security
Permissions in OneLake work at multiple levels. Workspace-level controls govern who can see a whole area. Item-level controls limit access to a specific lakehouse or warehouse. OneLake security adds fine-grained controls down to the table, column, or row, so a finance lakehouse can be shared with the wider business but salaries hidden from anyone outside HR. Sensitivity labels from Microsoft Purview travel with the data too, which means a piece of data classified as confidential stays classified when it is exported to Excel or embedded in a Teams channel.
What it means for your existing data estate
The single most freeing thing about OneLake is that it does not demand a migration. Most businesses we work with start by leaving the bulk of their data exactly where it is (Azure Data Lake, S3, on-prem warehouses) and use shortcuts to bring it into Fabric virtually. That gets reporting and Copilot working on the existing estate in days, not months. Migration becomes a choice you make later, system by system, based on cost and convenience rather than necessity.
Where to start
If you are evaluating Fabric, start by mapping the data sources you currently report on and which ones could live in OneLake natively versus which would be referenced via shortcuts. Our Microsoft Fabric and data readiness pages cover how we run that exercise with clients.