Super star clusters ( SSCs ) , likely the progenitors of globular clusters , are one of the most extreme forms of star formation . Understanding how SSCs form is an observational challenge . Theoretical studies establish that , to form such clusters , the dynamical timescale of their parent clouds has to be shorter than the timescale of the disruption of their parent clouds by stellar feedback . However , due to insufficient observational support , it is still unclear how feedback from SSCs acts on their surrounding matter . Studying feedback in SSCs is essential to understand how such clusters form . Based on ALMA and VLT observations , we study this process in a SSC in the overlap region of the Antennae galaxies ( 22 Mpc ) , a spectacular example of a burst of star formation triggered by the encounter of two galaxies . We analyze a unique massive ( \sim 10 ^ { 7 } M _ { \odot } ) and young ( 1–3.5 Myr ) SSC , still associated with compact molecular and ionized gas emission , which suggest that it may still be embedded in its parent molecular cloud . The cluster has two CO velocity components , a low velocity one spatially associated with the cluster and a high velocity one distributed in a bubble-like shape around the cluster . Our results on the low velocity component suggest that this gas did not participate in the formation of the SSC . We propose that most of the parent cloud has already been blown away , accelerated at the early stages of the SSC evolution by radiation pressure , in a timescale \sim 1 Myr . The high velocity component may trace outflowing molecular gas from the parent cloud . Supporting evidence is found in shock heated H _ { 2 } gas and escaping Br \gamma gas associated with this component . The low velocity component may be gas that was near the SSC when it formed but not part of its parent cloud or clumps that migrated from the SGMC environment . This gas would be dispersed by stellar winds and supernova explosions . The existing data is inconclusive as to whether the cluster is bound and will evolve as a globular cluster . Within \sim 100 pc from the cluster , we estimate a lower limit for the SFE of 17 % , smaller than the theoretical limit of 30 % needed to form a bound cluster . Further higher spatial resolution observations are needed to test and quantify our proposed scenario .