I don't think all 5,000 boards will be populated with chips this month. I think a more realistic estimate would be 2-3 months

3.5ph/s * 1w/gh = 3.5MW which seems like it would be more power than our little datacenter can handle (especially since Ken said he plans on hosting a maximum of 800th/s) thereby making most of the subsequent points moot.

As long as being a Hash Fast reseller isn't our endgame, I think ActM should maximize sales so as to build up capital for mass production of ActM's own chips which should carry a greater profit margin than acting as a reseller.