I find it interesting that at least some of the contributors to this thread seem to think that Arduino == atmega328. Personally, I include the entire set of Arduino compatible boards that can be used with the Arduino IDE + libraries. I bought an Arduino Uno years ago just to see what the fuss was about. Then I bought an Arduino Due (with cortex-m3) mainly because it had more RAM and I wanted to use it for data acquisition. Then I discovered Teensies. I have a half dozen Teensy 3.6 (cortex-m4) and just got a Teensy 4.0 (cortex-m7 @600MHz, 1MB RAM, 2MB flash.) I consider all of them Arduinos.
FWIW, I rarely use the IDE to write code; I use Notepad++ and use the IDE as a compiler. Is this a sophisticated and flexible programming system? Heck no. But it's entirely adequate for its intended purpose.
About me: I am an EE. Currently I spend about 1/2 of my time as an educator and the other 1/2 as a consultant. At my last full time job I designed embedded systems. I didn't program them though, and I typically let the software engineer pick the MCU. When I needed something in the lab with GPIO for control or data collection, I started by reprogramming an existing product; usually something with a 16bit PIC. Then I discovered Arduinos and never bothered trying to repurpose existing boards. I used them to control prototypes of other products that didn't include in on board MCU, or some that did have their own MCU but needed to be tested through some low level interface.
Again, I had piles of non-Arduino boards with analog and digital io at my disposal. All I had to do was program them. But when I found out I could spend $20 on a Teensy and cut my coding/debug time by an order of magnitude, I pushed all of those other boards into a box and put it on the shelf.
Did I or anyone else at that company ever consider incorporating an Arduino into one of our products instead of a $2 PIC? Of course not, that would be silly. Point is, I found good uses for both $2 MCU's and $20 Arduinos.
A project that I'm currently working on as a consultant involves reading a 3-axis accelerometer, doing a little DSP, then sending the result over Bluetooth. My first prototypes used the ADXL343 accelerometer (same that I will use in the final product) connected to a Teensy 3.6. With this I was able to perform both proof of concept and develop the DSP. There's only a few hundred lines of code and that include a command interpreter. The only thing missing is Bluetooth. For the commercial product, should I design in the Teensy coupled with more hardware for Bluetooth? Should I consider an STM32 because it's the new Z80? Maybe I should program it in Forth too...
No, I'll use a TI CC2640R2F. It includes a general purpose MCU, as well as a Bluetooth module, but the real reason I'll use it is because the tools and libraries available from TI will make programming the thing way easier than programming a PIC or STM32 from scratch to perform the same job. I expect that the time to write and debug the code will be reduced by a factor of 4 or 5. Hey, that's the same reason I used the Arduinos for prototyping!