What's the point of using instancing if i still have to cull faces? The reason im using instancing is because i could never figure out face culling lol

I think we're talking of three different types of "culling" here.

First, your cubes have faces that cannot be seen from any camera position because they are "congruent" with other faces. This does not change if your camera (player) moves, so this kind of "static" culling can be done once for many frames. Of course, you have to update the list of visible faces when you add or remove cubes (that is, you update the instance vertex buffer, but not the geometry vertex buffer or index buffer).

Culling faces that are visible from somewhere, but not from your current camera position is best done by the GPU. This changes from frame to frame as typically your player moves around in the world.

Frustum culling again needs to be done from frame to frame, but typically it is not applied to each single object, but to bounding boxes containing "chunks" of objects. So this can be done easily on the CPU.