Too much cDNA can be inhibitory (still contains various buffer components that are good for cDNA synthesis, bad for PCR).
Too much cDNA also results in inefficient amplification (you want your primers etc to be vastly in excess of your target, so reaction is limited only by [target]).
You do not need concentrated cDNA to get quantitative data (1/20 diluted cDNA will still allow you to measure everything from high abundance targets all the way down to extremely low abundance targets, and targets that require concentrated cDNA to even detect are clearly present at such low levels they cannot be meaningfully considered to be expressed).
It makes your cDNA last much, much longer: a 20ul reaction diluted down to 400ul, using 2ul at a time, gives you 66 triplicate reactions. Without dilution, you can measure only a fraction of that (and less accurately, because of 1-3, above).
Dilute cDNA 1/10-1/20 routinely, it is good practice.
About the only time I wouldn't dilute cDNA immediately is if I'm planning to use it for digital PCR, where you have a smaller dynamic range to play with: for some low abundance transcripts I use cDNA at less dilute concentrations (like 1/4).