When and where globular clusters formed? Did they contribute to the last major phase-transition of the our Universe? JWST wavelengths and sensitivity, combined with the magnification power of gravitational lensing is opening a unique opportunity to address these fundamental questions from direct observations. I will present recent analyses based on JWST observations of star clusters and stellar clumps in gravitationally lensed galaxies. While magnifications are not high enough to resolve these clumps down to pc scales, their physical properties are nevertheless important to understand where globular cluster progenitors form. I will discuss the physical properties of the young star clusters detected in the Cosmic Gems arc at redshift 10, recently discovered in the SPT-CL J0615−5746 galaxy cluster field. Five star clusters with sizes of ~1 pc and masses around 10^ Msun are detected in a few times 10^7 Msun metal poor and dust free host galaxy. These star clusters have elevated stellar densities and contribute at least to 50% of the FUV light and mass of the detected host. They have formed in a very compact region of about 40-50 pc. Cosmological simulations that focus on Milky Way assembly and formation of its globular cluster population find that massive star cluster formation takes place at redshift z < 7, thus too late to significantly contribute to reionization. I will discuss how the discovery of star clusters in the Cosmic Gems arc indicates that globular cluster formation in Milky Way progenitors (peaking at redshift ~3) might not be generally applicable, and star cluster formation and feedback could potentially be a necessary ingredient in galaxies that have contributed to reionization.