Scaling over the x axis means running multiple instances of an application behind a load balancer. This is a very common approach used in monolithic applications. One of the drawbacks of this approach is that any instance of an application can utilize all the data available for the application. It also fails to address application complexity.
Microservices should not share a global state or a kind of data store that can be accessed by all the services. This will create a bottleneck and a single point of failure. Hence, approaching microservice scaling merely over the x axis of Scale Cube would not be the right approach.
Now let's look at z axis scaling. We have skipped y axis scaling for a reason.