Notice that
$$ \begin{bmatrix} A & B \\\ C & I \end{bmatrix}\begin{bmatrix} I & 0 \\\ -C & I \end{bmatrix} = \begin{bmatrix} A - BC & B \\\ 0 & I \end{bmatrix}. $$
(This is nothing but LU factorization in block form.) Recall that a block triangular matrix is invertible iff its diagonal elements are invertible. (See, for example this.) Since $\begin{bmatrix} I & 0 \\\ -C & I \end{bmatrix}$ has diagonal elements $I$ and $I$, it is clearly invertible with inverse
$$ \begin{bmatrix} I & 0 \\\ C & I \end{bmatrix}. $$
Thus
$$ \begin{bmatrix} A & B \\\ C & I \end{bmatrix} = \begin{bmatrix} A - BC & B \\\ 0 & I \end{bmatrix}\begin{bmatrix} I & 0 \\\ C & I \end{bmatrix}. $$
The product of two matrices is invertible iff both are. The right matrix in the product is invertible iff $A-BC$ is invertible, as desired.