You actually could do it in either.
Generally you would model your object as a solid in the CAD program, and select the infill type and density in the slicer (Cura). This gives the most flexibility. 98% of the time, this is what you would do.
But if you have specific part needs, you could also model your part with the infill included in the CAD modeling, and then print with 100% infill configured in the slicer. This is beneficial if you need different fill patterns across the model, or have different strength requirements for portions of a part. For example, if you have a large "solid" mechanical part that needs a lot of material inside certain sections to avoid crushing from fasteners, but would be hugely wasteful of plastic and time to print with high infill across the whole part because other sections don't have any particular strength requirements.
The exact same situation is true with supports. For some models, it is far better to design them with built-in supports at the CAD stage, rather than relying on the slicer to auto-generate appropriate supports. By doing it yourself in advance, you have complete control over where the supports appear, their shape, where they attach to, etc.