Here's what Dan Lavry posted on his own web site 3 years ago:
"The Opamp approach has some advantages over discrete. The component matching capability is great, and one can match more then 2 devices, you can match dozens. Also, the ability to have everything packaged in a tiny area makes all the devices operate at the same temperature which further help greatly.
But Opams have their limitations as well, compared to discrete circuits. First, there are no inductors. Second, the capacitors used in the silicone are limited to very small values (the area is tiny). The caps are also limited in the type of material (fro example, no film type caps and so on…) Last but not least, most opamps are designed for relatively low current, thus limiting the applications when very high current is required. The same is true for voltage. Let me note that some companies specialize in manufacturing high voltage and current devices…
Some of the shortcomings of the opamps are overcome by using them with external parts. Often there are compensation pins on an opamp, enabling the designer to “continue” and “complete” the circuit inside the silicon with external parts. Often a designer can extend the usability of an opamp by use of external parts to cover high voltage, current and power applications, improve the input circuitry and so on…
But at the end of the day, the opamp is a repackaging job, and any comments against an opamp should be viewed as such. I am not for or against opamps. It is just one more part type, and one can and should use it were it makes sense. And one should not use it where it does not make sense. I would not “judge” the quality of a piece of gear based on use (or lack of use) of amps."
Andrew