A couple of months ago we analyzed historical microcontroller data to create the Octopart Microcontroller Price Index. We were excited to hear a lot of feedback on how to improve and augment the index from so many of our users -- thank you! This week we spent some time updating the index and examining some of the ideas and issues raised.
To begin with, we adapted our code to include new (post January 2012) microcontrollers in the index as they are introduced, a hole pointed out by many of our users. While the shape of the curves (pictured below) do change subtly, there is barely any appreciable movement. This is not unexpected; our sampling of parts by popularity naturally leans old. However, the original price index was certainly inaccurate without the inclusion of these newer parts.
In addition, we heard many calls to segment the data by spec. While gross category trends may reflect industry wide market forces, the circuit designer with particular needs will find data segmented by technical specs more actionable. To begin to address this, we segmented single component price trends by bus width, pictured below.
Unsurprisingly, the shape of each of these curves is not terribly different from the aggregate curve displayed above. In fact, the pricing trends for 8 bit and 16 bit microcontrollers look nearly identical. The most interesting feature to emerge in this analysis is the increase in 32 bit microcontroller prices over the last year.
As before, this feature begs many more questions than we are able to answer from the confines of our limited view of the industry. Are market forces pushing specialized 32 bit microcontroller prices higher and higher as 16 bit and 8 bit microcontrollers become more heavily commoditized? How is this affecting hardware developers in the wild? We'd love to hear your thoughts on these questions and, of course, more suggestions to improve our analysis!