The landscape in mapping the human brain has changed as far as the scale of data is concerned. Now that generation of data is easy, there is huge demand of solutions for (1) handling Petabyte scale image data (i.e. compression, untarring with bots from large-scale servers), (2) its 3D semantic segmentation with deep learning (i.e. automated approaches for volumetric reconstruction as manual approaches like crowdsourcing just don't work; Google is curently a big player in this working with IARPA and other labs across the globe) (3) development of visualization and analysis tools that allow web-based rendering of this PB-scale raw and segmentation data efficiently for neuroscientists/users (for example with React, js, etc. Many software companies have come up with solutions specific to users and their data formats).

As far as hardware developement goes, it is only specifically that would enhance these above software solutions. So really depends on which part of the pipeline interests you most in the above three (this is just a rough divison though).