Why haven't we got useful quantum computers yet?

Four years after Google first demonstrated the supremacy of quantum computers over ordinary ones, why aren't these exotic machines being used for practical problems?