My experience of paying for cloud services does not back that up. Java AWS Lambdas are typically 2-3x heavier in memory usage than a TS/JS equivalent, and the

My experience of paying for cloud services does not back that up. Java AWS Lambdas are typically 2-3x heavier in memory usage than a TS/JS equivalent, and the benefit of that warmup is negated if the Lambda shuts down due to idle time. "Give it up when asked" !== less streamlined, it means expanding heap is required for normal operation. Too little heap, it buckles, GC is almost constant, and performance goes out of the window.