Too much uninformed speculation here, so let's clear out the trash.
0) Start out by ignoring smartphones for a moment, because the power issues are significant enough that only in special cases (ie plugged in) will it make sense. Let's looks at other use cases first, specifically non-portable always on devices like routers.
1) 21s chipset technology will be trojan horsed into devices everywhere, at no additional cost to any customer. So you get the technology onto every device you have, courtesy of Qualcomm, Intel and Cisco, plus Chinese manufacturers for FREE. They can do this because adding some an instruction set to a chipset or ASIC is very near costless, if you only consider the marginal cost, ie if you're already building the chip for a router, adding an additional instruction set is a like a few cents of additional cost.
2) AirBnB for chips. The 21 team came from Counsyl, Counsyl came from Vijay Pande's lab. Vijay Pande, was the creator of Folding@Home. So you see, if you're paying 200 bucks for a router, which you only use the 6 waking hours of the day you're at home, then for 18 hours a day, those chips are idle. So that's unused computing power, which could be used for something like folding proteins... or mining bitcoins. Yes it uses power. But hey, you've already paid the capital expenditure for it. You're already paying the operating expense for the router while it is idle, ie vampiric power and uptime. So if you run the Bitshare chip, all you need to pay is for marginal power cost.
3) A lot of Bitcoin mining cost mathematics can get turned upside down if there is no capital cost. Ie if you don't have to earn back the cost of the hardware, then you only have to make more Bitcoin than electricity prices (and in this case, only more than marginal electricity use).
4) This idle hardware issue is actually a pretty large one. Remember that the Internet for the most part is built to handle surge or peak capacity. That's true of everything from data centers to Google. All that computing capacity is idle most of the time right now.
5)
Data center costs The cost percentages break down as follows:
Electricity: 20 percent
Engineering and Installation Manpower: 18 percent
Power and Server Equipment: 18 percent
Facility Space: 15 percent
Service and Maintenance: 15 percent
HVAC Equipment: 6 percent
Project Management: 5 percent
Rack Hardware: 2 percent
System Monitoring: 1 percent
Assuming utilization of 20% stepping up to 60% over a 10 year period with, average 40% utilization.
6) Based on 5), if Bitshare pushes up utilisation to 80% average, marginal electricity costs would go up at most by 20% of total costs. Meanwhile they've saved 80% of the cost of mining, which is the capital cost.
7) Remember Bitcoin mining is an embarassingly paralell problem, which means there are fewer economies of scale running it in a datacenter vs a router chip. So the analysis in 5) and 6) will hold at the router chip stage.
Remember also that data centers are already super efficient in terms of capital cost per compute cycle. When compared to consumer electronics, there is no physical distribution cost, no marketing cost etc.
9) In total, basically the marginal electricity cost is small enough that 21 thinks they can mine Bitcoins profitably enough, at current prices, as long as they don't have to bear the capital costs of the chips. In fact, they thing that it's going to be so profitable, that they plan to to take 75% of the Bitcoins mined, and leave you with 25%.
10) That 75% of Bitcoins mined is going to be distributed back to their partners, ie Qualcomm, etc for profits. If this is a profitable business for them, they will start packing in more and more powerful chips into consumer devices. Because utilization of the chips is going to go from 20%-60% to 80%-100%.
11) Hardware as a Service. You pay electricity costs for the mining, and Qualcomm subsidizes the chip costs, over time capital costs for the chips will come down because someone else is paying for it.
12) Over time, 21 only makes real money if Bitcoin becomes huge. They want to be a payment processor, they need volume of transactions. They are not interested in holding the Bitcoin as an investor.
13) So that leads to the other part of the business plan, the 25% share to you. That share, which is basically "free money", allows developers to provide interesting and innovative applications on the Bitcoin platform. Many of these applications may be at first machine to machine applications.
