I did a master's thesis on visual inertial navigation just a few years ago. Around the time when ARKit and ARCore got released (what a time to be robotics researcher). At that point global shutter cameras were significantly more expensive. Basically you had to go for what is called "machine vision cameras". They were very robust, deployment ready and highly capable and configurable. A camera with similar parameters to this one could cost you anywhere in the neighbourhood of 1000 - 3000 USD.
It is great to see that they are getting very affordable. I think this trend will help bring visual or visual-inertial navigations to the arsenal of average makers. Before, you had to have a very advanced software to compensate for imperfections of your sensors and the calibration complexity meant that only professionals could have them running reliably. We will se what people make with them.