Well for a start your second piece of code would return Task<T>
rather than T
. The ultimate answer is "it depends".
If your page needs to access multiple data sources, it may make it simpler to parallelize access to those sources, using the result of each access only where necessary. So for example, you may want to start making a long-running data fetch as the first part of your page handling, then only need the result at the end. It's obviously possible to do this without using async/await, but it's a lot simpler when the language is helping you out.
Additionally, asynchrony can be used to handle a huge number of long-running requests on a small number of threads, if most of those requests will be idle for a lot of the time - for example in a long-polling scenario. I can see the async abilities being used as an alternative to SignalR in some scenarios.
The benefits of async on the server side are harder to pin down than on the client side because there are different ways in which it helps - whereas the "avoid doing work on the UI thread" side is so obvious, it's easy to see the benefit.
Don't forget that there can be a lot more to server-side coding than just the front-end. In my experience, async is most likely to be useful when implementing RPC services, particularly those which talk to multiple other RPC services.
As Pasi says, it's just syntactic sugar - but I believe that it's sufficiently sweet sugar that it may well make the difference between proper asynchronous handling being feasible and it being simply too much effort and complexity.