This can seem like a no-brainer initially as it would seem that everyone has JS enabled. The percentage of users online with JS disabled is a mere 2-3% depending on where you look. JS functionality is also supported on the mobile platform which is growing steadily. And besides, how unexciting would it be to traverse the web sans JS - imagine all the cool functionality you would miss!
All that aside though, site owners can sometimes only gleen analytics on users with JS enabled to improve the user experience. But, the reality is that some users just don't like JS for privacy reasons or they don't have JS support because they have an older or non-JS browser, etc.
However, much of it comes down to knowing your users. How many of them have JS disabled or are on non-JS supported browsers? The percentage might be higher or lower than the worldwide web average and the actual number of users may or may not be significant. Also, it depends on what locales the site serves as the number of users disabling JS varies even within the US and globally.
It is interesting to note that JS usage has been trending upwards. As well, the latest version of web markup, HTML 5, is supposedly able to achieve even more powerful capabilities and user experiences when paired with JS.
If at the end of a cost-benefit analysis it's decided not to create a non-JS version, what is the opportunity cost associated with this and is it significant? It's clearly ideal to plan for this from the outset and incorporate support for non-JS folks, but I don't think anyone would argue with focusing on creating the best possible experience for the majority of your users and working your way towards any edge cases.