When all costs are considered, electricity from an on-board battery is a lot more expensive than electricity from a wall outlet, and absurdly more valuable for the owner of the device. Therefore, placing miners on mobile devices like smarthphones, tablets, or laptops makes no economic sense.
While a mobile device is being charged the electricity cost is equivalent to wall outlet costs. Many users of mobile devices like laptops use them while they are connected to an outlet. The mac book pro I am writing this reply on now is a good example.
Moreover, I cannot imagine any advantage of phsically attaching a mining device to some other device with a different function, except water or space heating. There is a thing called "internet" that allows a device to send information to another device anywhere in the world; and a system called "bitcoin" that is supposed to allow bitcoins to be sent through the internet, even if the two parties are not connected to the internet at the same time.
I can easily imagine Qualcomm including 21 inc's proprietary IP in new chip designs.
So, instead of having 10 mining chips scattered around, firmly attached to my devices, I could have one mining rig with 10 chips sitting anywhere it is more convenient (for example, in my water heater, or in the attic of my summer home in Iceland), sending bitcoins to other devices if, where, and when needed. The centralized rig would certainly be more efficient, could work at steady pace 24/7, use optimized power supply -- and would not burden the other devices with power load, extra heating, board space, etc.. Physically attaching a mining chip to a router makes as much practical and economic sense as attaching a water filter to a toaster, or a lawn mower to a car.
This is obviously an adoption play. No one is likely to come up with a workable business plan to place dedicated miners each containing 10 chips in the attics of 10 million people, but it is easy to conceive of 100 million routers/USB hubs/phones/devices each containing a chip with 21 inc IP included on die at near zero incremental cost.
I cannot make sense either of the fuzzy idea that, with miner chips, devices would be able to get revenue to do micropayments for the services they use, or to generate satoshis for "Wide World Ledger" applications like colored coins, authentication, timesamping, etc. It will be impossible to exactly match the bitcoins produced by a device to its needs: if the device does not run out of bitcoins, it will generate a surplus of them. My water heater could generate a lot more bitcoins than my laptop, but will hardly need them; whereas my laptop wil probably will need them the most.
So, if my devices had embedded miners, they would have to pool their bitcoin streams and share them on the basis of need; and possibly topped up with purchased bitcins. Which, again, is better done with a centralized mining rig than with independent miners. (By the way, in most higher organisms, from corporations to computers to humans, evolution has found it better to use specialized units for the generation of "revenue" (money or energy), separate from units devoted to other functions that spend from that "revenue".)
When there are satoshi to be spent, people will figure out how to spend them.
Finally, adding a mining chip to a battery charger makes it much more complicated, functionally. Instead of just taking AC 120 V and putting out DC 6 V, a "miner-enhanced" charger will have to communicate with 21.co servers, manage a wallet, interact with other devices, authenticate itself, etc.. Even if most of that complexity is hidden from the owner, enough of it will be exposed to require some of his attention and time. Which are much more expensive than the electricity or the chip itself...
There is already an enormous amount of complexity hidden in the simplest of everyday devices, and orders of magnitude more in modern mobile devices. This is no challenge.
