Recurrent glioblastoma has no established standard of care, and survival remains limited to a few months. Recent clinical evidence suggests that the intracranial delivery of CAR-T cells is feasible and shows signs of antitumor activity. In the phase I trial NCT02208362, anti-IL13Rα2 CAR-T cells were tested in 65 patients, yielding a 50% disease-control rate and 23% one-year survival. Two phase I studies of multi-target CAR-T products (NCT05168423, the ongoing NCT05660369) similarly showed bioactivity and early but transient radiological responses. Importantly, the intracranial route enabled rich translational research: serial biopsies and cerebrospinal fluid sampling allowed assessment of CAR-T cells function at the tumor site and the identification of candidate predictive biomarkers. In this review, we compare the designs and findings of these three studies and synthesize key lessons. We also outline future perspectives to improve the efficacy of CAR-T cell therapy in solid tumors, with a focus on glioblastoma.