CPU usage is often a red herring because you don't know if it represents useful work or not. If you have high CPU usage, it could indicate that the application is doing a lot of useful work (good); on the other hand, it could indicate that the application is being very inefficient (bad).
It's better to focus on direct measures of useful work, such as the throughput and latency of the application. For a web server, these would include metrics such as maximum number of simultaneous users and the time taken for the server to respond to each request. Then you can compare these metrics to CPU usage to get an idea of CPU efficiency.
To answer your specific question: it depends on the performance bottlenecks in your particular environment.
In many situations, adding RAM will actually increase the CPU usage because it will allow more useful work to be done -- and this is a good thing. For example, imagine a web server under heavy load serving up files. The performance bottleneck is likely to be disk I/O; the application will spend most of its time waiting for disk, so the CPU usage will be quite low. If you added more RAM to cache the files on disk, the application would spend less time waiting, more time processing, and so CPU usage and throughput would be higher.
It could also work the other way. For example, there is often a trade-off in computer algorithms between space complexity and time complexity; that is, extra RAM allows the application to use less CPU, so CPU usage would be lower (or you would get more throughput per CPU cycle). @sybreon has given a good example of how adding RAM can make the CPU usage more efficient because the system doesn't have to spend as much effort managing a constrained resource.