Actually memory accesses are much slower than CPU speed, memory runs at about 150ms or (in newer systems) at 333ms or so. But current CPUs are equipped with a on-chip memory cache which works at CPU clock speed and which is from 512 KB to 2MB so the computations mentioned here are correct. Also second cache is placed between the CPU and the main memory which runs at about 1 GHz and can store up to 16MB so the slow memory access impact is reduced.
Also there exist systems that are running all their memory at CPU speed. These systems are very expensive but they do exist and for them these computations also work.
For normal systems with not-so-large cache you should realize that a typical modern operating system contains over 10000 various binaries which have average overhead of several tens of kilobytes so the computatuion is also valid for them (the memory is only about 1000 times faster than the harddisk but there are ten times more programs and ten times more overhead).