Prototypes cost money and time. Designing on a Computer costs only time in a seat for a person / people and designing in software is great because you can try many things very quickly. Sadly what this tends to mean is that in a lot of companies where hardware is not the main focus, there is a hesitation to build a prototype because it is a cost that the teams are sometimes not built for. Sometimes the hesitation can come from not knowing what type of prototype to build or what that prototype will cost before a lot of other work is done first.
There are however many challenges that require building something physical, sketching something down, or testing something out on people because the risk associated with it is very difficult to deal with sitting at a computer or in design meetings, brainstorming sessions etc.
The more you products you develop the more comfortable you become with this process, at the same time as you get comfortable with when and what types of prototypes to build, you are also building expertise and reducing the number of prototypes you require as you have already addressed some of the more standard hardware design problems that exist in the world. Waterproofing, Lighting / Light Piping, Audio / Sound Quality, Ergonomics, Tolerance Analysis, High / Low / Fast Changing Temperature Constraints, Motion Analysis, Mechanism Planning, Force Analysis, Design for low / medium / high volume production, Maintenance, Repairability, Drop Testing, etc. Sometimes the challenges come when you need to address many of the above all at the same time and competing goals make individually addressable items more complex and the fastest way to deal with it can be a fast prototype to address the risks.
At Design 1st we make 2000+ prototype parts a year, but we are always looking at how we can bring elements of physical prototyping into the digital realm so that we can iterate faster and reduce risk and design cycle time. At a minimum we build a Production intent prototype at the end of the detailed design stage when we are getting ready to head to volume production. The reason we insist on this prototype is to address all the decisions that have been made in the design process, and to check final issues before spending much more significant money on the equipment required for mass production.