Some hardware producers believe that, while agile methods are important for developing software products, they do not apply to hardware products. In my view, agile methods benefit development of all types of products.
As I mentioned in a previous entry, the benefit of agile development is that you build into your process an efficient means of dealing with mistakes and discoveries that inevitably transpire. You discover requirements you hadn't considered. You find inadequacies in the design. You want to make these discoveries as early as possible in the development process. Iterating on the product development tends to minimize the amount of time it takes to discover problems and inadequacies and recover from them.
Why might agile processes not work for hardware development? In some cases, the cost of a hardware development iteration is much higher than that of a software development iteration. In software, your development staff simply updates code in each iteration. Code costs nothing but the time to write it. The equipment in hardware, on the other hand, costs money above and beyond the time to fabricate and piece it together. If you iterate too much, these costs will add up and exceed the savings you derive from discovering, and adjusting to, mistakes earlier in development.
The cost of iterating clearly does impact the wisdom of using agile methods. However, only in extreme cases does it ever render agile methods inferior to waterfall methods. Furthermore, organizations with no experience with agile methods tend to overlook the many creative ways they can iterate at a low cost. This myopia tends to bias them against agile methods.