This study investigates the implementation of Graal Virtual Machine (GraalVM) in Java Spring Boot microservices for mobile banking applications. Increasing digital banking demand and user growth necessitate systems that can handle high transaction volumes efficiently. Java Virtual Machine (JVM) environments face challenges, including slower application startup times, higher CPU usage, and increased memory consumption, limiting their suitability for such high-demand scenarios. To address these issues, this research uses a quasi-experimental design to compare microservices' performance on GraalVM and JVM by analyzing key metrics: application startup time, CPU usage, and memory consumption under various load scenarios. Results show that GraalVM significantly improves startup times, reducing delays by 25–30 seconds across services, thus enhancing responsiveness. CPU usage showed varied outcomes: mobile-service demonstrated reductions (e.g., 0.8678 to 0.7798 for 100 users), whereas profile-service and casa-service recorded slight increases under certain workloads (e.g., 0.7829 to 0.8569 for profile-service at 100 users). Memory consumption increased notably for GraalVM, particularly in high-load scenarios such as casa-service at 600 users (198.47 MB to 591.38 MB). These findings highlight the trade-offs of adopting GraalVM, with faster startup times offset by higher memory usage in specific services. The results underscore the importance of workload-specific evaluations when optimizing microservices. Practical applications of this research include guiding system architects in selecting appropriate runtime environments to enhance the performance and scalability of mobile banking systems, ensuring efficient operation under varying demands