In the standard galaxy formation scenario plasma clouds with a high thermal energy content must exist at high redshifts since the protogalactic gas is shock heated to the virial temperature, and extensive cooling, leading to efficient star formation, must await the collapse of massive haloes (as indicated by the massive body of evidence, referred to as downsizing). Massive plasma clouds are potentially observable through the thermal and kinetic Sunyaev-Zel'dovich effects and their free-free emission. We find that the detection of substantial numbers of galaxy-scale thermal Sunyaev-Zel'dovich signals is achievable by blind surveys with next generation radio telescope arrays such as EVLA, ALMA and SKA. This population is even detectable with the 10 per cent SKA, and wide field of view options at high frequency on any of these arrays would greatly increase survey speed. An analysis of confusion effects and of the contamination by radio and dust emissions shows that the optimal frequencies are those in the range 10-35 GHz. Predictions for the redshift distributions of detected sources are also worked out.
