The chip is crazy and amazing.
pics or it didn't happen 
*picture of hashrate graph*.
Seriously guys, use some sense. They posted the first real life test results from an ENGINEERING SAMPLE and suddenly everyone is claiming they just deployed 100PH of them.
It wasn't my intention to claim such. I'd love to buy bitfury stuff again, but we are not in 2013 anymore and the chip has to be really incredible to make sense to sell it to the public.
Bitfury chips have been that way since gen1.
rev1 (55nm)(Sept 2013): 0.9-1.2w/GH, improved slightly by the first ASIC string for bicoin mining. Sold as a barebones miner that was easy to produce and could work without heavy/costly heatsinks. about 2x the efficiency of the competitors
rev2 (28nm)(Feb 2015): 0.2-0.3w/GH (4x gain), pretty much exclusively private deployment. about 2x the efficiency as the competitors
rev3 (16nm)(Feb 2016?): 0.05-0.07w/GH (4x gain), hopefully they sell to public
these guys have been the smartest group so far IMO, opting for high-density designs that run in string (less DC/DC components) and dont require heavy heatsinks, only a sparsely populated PCB. deployment to an immersion system would be fast for hardware like this, and the string design could allow operation from >12V power. I would not be surprised to see thier in-house hardware operating with 48VDC power supplies for density