Introduction: The current the learning curve for robotic pancreatoduodenectomy (RPD) is described to be 20-40 cases. This study aims to characterize trends in RPD over time, associated patient outcomes and opportunities for advanced trainees. Methods: Using the ACS-NSQIP database from 2014-2019, PD cases were studied, and operative approach (open-OPN, laparoscopic-LAP, robotic-ROB) was characterized. Postoperative outcomes were described by year and operative technique. Results: During the study period, a total of 24,268 PDs were identified, 71% of which were for malignant disease. The ROB approach increased from 2.8% to 7.5% while the LAP approach decreased from 4.2% to 3.6%. Unplanned conversion increased over time for LAP (27.7 to 40.4%, p=0.003) but was unchanged for ROB cases (14.8% to 14.7%, p=0.257). Morbidity increased in OPN PD (35.5% to 36.8%, p=0.041) and decreased in ROB PD (38.7% to 30.3%, p=0.010). Mean LOS was lower in ROB than LAP and OPN (9.51 vs. 10.90, p<0.00001). In 2019, only 360 RPDs were performed in NSQIP, which accounts for approximately 70% of the PDs. Given a total of 100 AHPBA, SSO, ASTS fellows, only 5 RPDs are available per trainee per year which is far below the learning curve. Conclusion: Over a six-year period, a significant increase was observed in the use of RPD without a concomitant increase in conversion rates. RPD was associated with decreased morbidity and length of stay. Despite a shift towards more RPDs, the number of cases being performed in North America is not adequate for all fellows to achieve the learning curve before graduation.