Part of the challenge in engineering is to know when to use certain processes or devices and when to use others. For example, a tube in the input stage works well, but not so much in the output. Or, a capacitor used as a DC blocker might sound better than the complexity of a servo, or, vice versa.
Analog integrated circuits, like op amps, can typically be bettered by their discrete counterparts in some cases, but not all. For example, if component matching is a critical aspect to your design then there's likely no better process than integrating everything on a single piece of silicon. Each component tracks the temperature variations of the other for near-perfect matching.
Yet, in the same way separates can outperform integrateds, there are disadvantages to IC solutions too. The limitations of single silicone, including low power requirements and a lack of isolation between components, can hinder performance levels in highly resolving systems like the kind you and I might want at home.
It's always a good idea to keep sweeping proclamations of better and worse at a minimum.
Like just about everything else in life, it's the details that flush out the devil.