Agilent has oscilloscopes to 63 GHz /160GS/s. How is that possible?

There is a white paper on this scope which describes a 10 ns clock with really clean edges - sub-sub 1 ns risetime.