rka,

Your strategist is quite right that a dimensional model is not a requirement or a complete solution for a data warehouse. Dimensional models are designed very specifically for data marts supporting an OLAP-style presentation tier (such as SSAS multidimensional) but they are less well suited to things like ad-hoc analysis, dashboards, "static" reports and downstream data integration. A normalized, atomic and subject-oriented data warehouse serving as a hub for dependent data marts is a more versatile and scalable approach than dimensional marts on their own.

A normalized data warehouse also reduces time and cost to develop, support and maintain the data warehouse environment because it ensures that presentation tier requirements are decoupled from the data sources themselves and doesn't suffer any inbuilt "bias" for a particular pattern of current requirements.