Actually it really depends on the quality of the design for the modules and it also depends on the features/performance of the ICs coupled into the business model (volume/quality metrics) of the industry it is intended for.

For example, there are now many good RF modules for Bluetooth and for Point-Of-Load (POL) switching power modules like those from TI and Artesyn. The RF modules are FCC approved and if you have ever gone thru an RF approval, then you would appreciate the time, costs, and complexity of that.

In addition, the POL modules are fine tuned for mid to high 90% efficiency, so wasting design time to brew your own is not only costly but inefficient. This is also true for many other types of power supplies and some such as medical device approvals are excruciatingly hard to approve.

For a specialized full custom design where the company is worried about saving pennies AND will be producing MILLIONS, then you must brew your own to save the company money (e.g. consumer and automotive markets)

For many other industries such as Military, Space, and Defense, there must be very specialized parts that most likely are not readily available from many sources and may require a lot of custom support. In these cases it is best to find a 'close' module and see if the vendor can customize it for you. Otherwise building circuits from ICs may not be an option; especially if it does not exist or has never been done before, as is typical with cutting edge Military and Space programs.

So there is no such thing as better, but rather what makes the most sense based upon what's available, what the price is in a specific quantity, and the time/effort trade-off of doing (brewing) it yourself.

Thanks for the reply Steven,

This question has been playing on my mind.
There seem to be quite a lot of big advantages in using modular design in many situations including large benefits regarding recycling however as you say there are situations where integrated design is vital or essential especially in high frequency design.

Also this question can be applied to building your own circuit or buying a or evaluation board, demonstration board, or development board.

An evaluation board is usually just a fan out of an IC with very little to no support circuitry. It is used to evaluate (typically) the functional pins IC itself of WHAT the IC is supposed to do as features.

A demonstration (demo) board is a board that is used to demonstrate the functionality of the IC and literally has all the bells/sounders, LEDs, motors, sensors, or other functional ICs that demonstrate HOW it can be used.

A development board is something that any good embedded programmer knows what this is. Its a board that has all the interface pins broken out with a JTAG interface to be able to program it repeately. It also has some support LEDs, connectors, oscillators, etc to be able to connect the board into a system. This is the tool of choice to get the software engineer going when the hardware engineer is still in the schematics phase. It is a vital tool for any project and applied to any embeded device such as MCU, CPU, GPU, DPU, DSP, FPGA, SoC, MP-SoC, RF-SoC.

So does the engineer prefer to use any of these tools or dive right in and create a custom board immediately ?

For designs that have high confidence and are evolutionary, usually there is no need to prove out or integrate many eval or dev boards. An engineer usually just wants to get it done as fast a possible.