Every simultaneous request is in its own microVM, making concurrency a non-issue / thing to optimize in most cases at the function level.