30GHz is way up there for a prescaler, and ones that will run
that fast are probably not "digital" per se, but RF parts that
either presume a ground referred and symmetric, or a capacitor-
blocked RF input. The divider element will have to be a dynamic
flip-flop, normal digital CMOS ones would need a technology raw
gate delay in the 5-7pS range to have a prayer of toggling
at the upper clock rate.
You might take a closer look at the Hittite part's specs and
apps stuff. Ratings for RF prescalers often have some funny
dependence on input power, more power can go faster but
not everyone is willing to throw +20dBm at it; to make it RF
market friendly, maybe they spec at 0dBm input power and
leave some upside on the table for you to pick up? Similarly
I have seen parts which have a bias feed resistor and can
change the bandwidth a fair bit if you're willing to burn the
power.
I'd also peer at that low end of 10GHz, seems kind of a tight
notch of operation. What makes it so, can you just use a
bigger (but then higher ESR/ESL, and more insertion loss
before the clock pin, meaning you have to again throw power
at the problem) blocking cap, or something?