I don't think microarrays are dead, not yet, for a number of reasons.
Despite what WouterDeCoster says, I think they are still cheaper than RNAseq, allowing you more replicates for the same amount of money (so more reliable measurments).
It is definitely true that the landscape of what you profile is limited to what you spot on the array, but then for most applications people just collapse all reads mapping to a given gene even after doing the fancy RNAseq stuff.
I also think that analysis of RNAseq data is still much more challenging, in terms of computational resources as well as preparation of the analyst. Microarrays have been around for so long that there are many many tools to analyse them without much prior preparation. Space and RAM-wise, well: I can easily analyse from start to end a relatively large microarray dataset in my home laptop, but I definitely cannot do this for a large RNAseq dataset.
Here is an example of a recent publication in which expression profiling is microarray based: http://www.nature.com/nature/journal/v535/n7612/full/nature18637.html.
There are many more: not dead yet.
Based on what I know about prices a microarray is not cheaper than RNA-seq (depending on desired number of reads of course), and RNA-seq is also more sensitive, giving you nucleotide level information and information about unannotated genes/transcripts. Therefore, I don't see a reason to use microarray over RNA-seq.