Introduction: Carbapenems are frequently initiated empirically for suspected neonatal nosocomial sepsis, driven by concerns for resistant Gram-negative pathogens. However, the extent to which empirical carbapenem use aligns with actual microbial ecology and resistance patterns remains unclear. Understanding this ecology and prescribing interface is critical to optimizing stewardship without compromising patient safety . Objectives: To evaluate the relationship between empirical carbapenem prescribing, microbial ecology , resistance patterns, and stewardship review outcomes in a tertiary NICU. Methods: We conducted a retrospective cross-sectional utilization and stewardship outcome analysis over a one-year period in 2025, involving 46 neonates initiated on meropenem for suspected nosocomial sepsis. Data sources included days of therapy for meropenem utilization (DOT/1000 patient-days), empirical versus definitive therapy , 72-hour antimicrobial stewardship (AMS) review outcomes, organism distribution, and NICU antibiogram. Results: Total meropenem utilization was 37 .8 DOT/1000 patient-days. Empirical therapy exceeded definitive use (22.8 vs 14.9 DOT/1000 patient-days), with a 1.5-fold higher exposure. Median empirical DOT surpassed definitive DOT (22.6 vs 9.2). Among 46 reviewed cases, 41.3% were de-escalated by 72 hours, while 52.2% continued therapy . Culture-negative clinical instability accounted for 75% of prolonged courses. Carbapenem exposure was predominantly front-loaded prior to microbiological confirmation. This reflects ecological pressure from Gram-negative nosocomial pathogens led by Klebsiella pneumoniae. Conclusion: Diagnostic uncertainty remains the principal barrier to early carbapenem deescalation, highlighting the need to strengthen culture yield and targeted empiric algorithms. Integrating ecological surveillance with structured AMS pathways is key to optimizing carbapenem stewardship in NICUs.