It is more complicated than that. You can look at it like this; M is more part of the ETL process of preparing, cleaning and merging your data, while DAX is more about the insights.

If you can, it is always better to use M (the rule is to make modifications as close to the datasource as possible). You should (mostly) use DAX to create dynamic insights, so for example numbers that should be calculated on the spot based of the user's input (slicers etc).