If you have two linear equations,

and

then we can represent the system of linear equations with a matrix equation.
Now if we were to try and solve this matrix equation for

(which represents

) then we would try to first take the inverse of

and multiply both sides by it, since we are trying to get

on it's own (we need to get rid of the

on the LHS side, and we can do that by 'hitting' it with it's inverse from the left to give us the identity matrix

, which effectively leaves us with whatever matrix we multiply that by).
Now this will only work if we have a unique solution, that is we can only find an inverse of

if the system has a unique solution. For an inverse of

to exist, the determinate of

, or
)
must be non zero.
So this means for:
- a unique solution,
\neq0)
- no/infinite solutions,
=0)
For no and infinite solutions case you will then need to be able to pick between the two by substituting your values found back into the equations and seeing if the come out as the same equation or not.
Hope that helps

EDIT: tl;dr, it comes down to the fact that you need the inverse of

to exist or not exist, which is then linked to the determinate of the matrix being zero or non-zero.