Data lakes provide a better end-user experience, are inexpensive to maintain, and require no additional engineering resources to construct.
The primary benefit of data lakes is usability for the business. The middle layer of the data ingestion machine; Analysts or Business Intelligence Engineers, were once required to interpret complex data models to deliver value to the business, the data can now be connected directly from source to end-user. Analysts and Business Intelligence Engineers can now focus on solving more valuable problems, such as engineering features to build predictive pipelines. The recent success of data lakes show that compute and storage resources that no longer benefit from being marginally reduced, and increased usability has been re-discovered as a major overall uplift to the data ecosystem. The maintenance cost associated with dimensional models could instead be dedicated to creating rapid-value for the business.
What about dimensional modeling? Dimensional modeling has its time in history, and much like the cube, I believe it will drift into obscurity. There are many companies today that are deeply committed to dimensional modeling, so I don’t believe the skill will die out for many years. As new teams begin evaluating the cost of data lakes and dimensional models, fewer and fewer dimensional models will come to exist.