Why can't we use data warehouse directly instead of database?

Question

I am learning the Hadoop and came across the HIVE. And I learnt that HIVE acts a data warehouse for analytical queries.

If data warehouse is aggregation of multiple data sources, why do the companies use multiple data sources and aggregate them? Why can't they directly write to data warehouse as it would reduce the overhead of aggregation and processing?

Is it because the overhead that would occur if all the users read from same data warehouse?

I wanted to learn why do we need databases if we have data warehouses which are highly scalable?

(Only in context of big companies that generates huge volumes of data. Because I can understand that most of the companies can even manage with 2 databases.)