Lets ponder for awhile how feasible this approach could be.
What are the current rates?
As with all analysis related to Bitcoin mining and profitability, everything is in the state of flux. So here are some current numbers we will be using for those of you who might be reading this in the future.
- Current Bitcoin block reward - 25BTC / block
- Current difficulty - 48'807'487'244
- A representative small-form ASIC miner - SINOSIGMA 3.6 GH/s for $25, containing one A3233 Chip, consuming ~0.75J/GH and costing about $1.5 per chip
- Current Bitcoin price - about $234
Based on that, we can do some calculations. That chip is currently mining about 0.000037BTC per day, earning about 0.013540 BTC or $3.17 per year if you run it at no cost. If you take the cheapest electricity cost of say, India or China at $0.08 / kWh, your annual profit is about $1.28, so you break even for the cost of a single $1.5 chip after 428.5 days of mining.
The current standard transaction fee in Bitcoin is about 0.1mBTC per 1000 bytes. The chip could make this much in about half a week.
Currently the difficulty on the Bitcoin network has slowed down its growth. It currently is about 48B, 5 months ago at the start of the year it was 40B, it was 23B in September of last year, and 13B in July of last year. So the difficulty has grown by roughly a factor of 4 since last year, but currently it looks like it slowed down to under a factor of 3 annually.
Estimating how much bandwidth a Bitcoin miner takes can be a bit tricky. Some people have reported it using about 20MB per day.
The difficulty approach, versus the simple approach
Looking at all those numbers, it looks like the mining doesn't scale all too well. In order to mine a dollar of bitcoins using a single chip it would take a phone non-stop 4 months of work. During that time, the phone would need to be connected to the Internet, powered up and heating up from the mining. Subtract from that the cost of electricity, mobile data and so on, and you might be operating at a loss in most situations. This is not to mention the decreased battery life from higher heat exposure and so on.
All in all, ASICs operate the best with economics of scale in mind - producing chips in bulk, cramming them into well-ventilated chassis, putting those in a data warehouse and so on. Perhaps instead of cramming a cellphone with an ASIC chip, it could come with some cloud mining subscription for life where a given datacenter would mine for BTC in the device's name and send those coins to the address bound to the phone over the years?
Alternatively, do away with mining and just buy some coins up front. Add $5-$10 to the phone's price up-front, but BTC with that money and trickle the coins into the device say, once a week or whenever the balance is getting low. Heck, if you notice a device's balance is still positive, you can always keep the coins for longer and perhaps even cash the coins after a year of inactivity of a potentially dead phone. This solution is much simpler and elegant. Best of all, since all Bitcoin balance data is public, you can top people's phones up without checking with them - you will know when their balance is running low.
Sometimes the first solutions might be the most obvious, but there are more elegant solution to solving the same problem. Cramming Bitcoin mining into devices in a way that detracts from their primary usage doesn't make sense. If you want a device augmented with Bitcoin mining, make sure its primary function is generating heat - this way even if the toaster or the space heater doesn't earn any money, it will still be a useful source of heat, rather than a hotplate for a phone that just eats your bandwidth.