There are many different reasons for many different languages.
MEMORY
When you declare a variable, you want some memory to hold the variable in. This involves asking the kernel of the operating system for memory, or some kind of monitoring program which keeps track of memory. In short, this can be an expensive operation.Hence, in many cases, it is desirable to allocate all the memory required for the object at the same time, and then assign whatever value has to be assigned to it later. This way, you can increase the performance of the program in the critical parts. This use case is common enough that a feature allowing declaration without initialization is allowed. However, good practices assert that in all other cases you should initialize the variable while assigning.
Think of the memory allocation as a bureaucracy. There is too much paper work. So, if you know you are going to use a large amount of memory later, you ask for a large amount of memory upfront in one single transaction, rather than asking the kernel each next time.
EXPENSIVE INITIALIZATION
This point is very similar to the above point. Suppose you have a 1 million times 1 million array. Initializing such an array is an expensive procedure. To do so with defaults would be stupidity, and hence, such a feature, where memory is allocated and then used as needed.
In here, its like you are buying a huge amount of lego blocks to construct something, but you want to buy them in shapes of the default spiderman. The shopkeeper or you would have to extra hard to get them in shapes of spiderman when you are anyway going to reshape them later.