I would have thought the hard economics of Bitcoin mining would pretty much instantly rule out of contention the politically correct renewable energy darlings wind and solar as viable sources of the computing power—i.e., electrical power—that underpins the world’s biggest cryptocurrency. Miners make money when the cost of the electricity that processes the many billions, sometimes trillions or quadrillions, of calculations per second, plus the cost of their mining equipment (advanced single-purpose processors and ancillary electronic paraphernalia) amounts to less than the value of the Bitcoin rewards they earn when the blocks of transactions that they have validated through sheer brute computational force are accepted as part of the global public electronic ledger called the Blockchain.
A miner makes or loses money depending chiefly on two things. First is the value of Bitcoin at any given minute. This value has been fluctuating significantly though trending upward in recent months; see the chart below.
Second is the level of difficulty in finding the “target,” i.e. the 256-bit number that is equal to or less than the encrypted hash of a block’s header. Finding the target is what puts the block into the blockchain and earns financial rewards in the form of Bitcoins. Note the 256-bit target size, and the word “encrypted.” This means that finding the target is, by design, not easy: until somebody figures out how to crack the encryption algorithm that Bitcoin uses (SHA-256) there really is no other way to find the target than to increment the nonce, cross your fingers, and compute the next hash.1
Using a normal desktop computer, finding the target to solve a Bitcoin block could take up to 140,000 years, according to the Princeton University Bitcoin and Cryptocurrencies lecture series. (You can get the series book for free here. The viability of CPU-based mining is discussed on p. 138.) Obviously you will need something a lot faster than a desktop computer.
If you had a stack of say 100 graphics cards hooked up in series you could drop that estimate down to 140 years.
… in the case of solar power, when the electrical watts stop altogether, as they do every evening when the sun goes down… well, do I even need to say it? Sadly, from my experience with solar enthusiasts, yes I do need to say it.
But since rent is due each month, not in the year 2157, you wouldn’t use a stack of 100 or even 1000 or even a million GPUs. You would use purpose-built Bitcoin mining ASICs (Application Specific Integrated Circuits). And you would use a bank of them, because the number of possible target candidates is something in the order of 267. This is a huge number: more than 147 billion billion. It is much, much easier to find a needle in a haystack.
Bitcoin miners are solving blocks and having them validated on the Blockchain every 10 minutes or so. The total electrical power required to make this happen is estimated to be something on the order of 900 megawatts.
Now, if electricity were free, then mining economics might be a step closer to making the astronomical odds against solving a block a bit less daunting. “Free” electricity from solar panels is the basis for much of the ridiculous greenwash from solar energy enthusiasts who have glommed (as I guess I should admit I have) onto Bitcoin as, if not quite the very future of money and all transactions, then at least a very fascinating phenomenon worthy of attention.
The prevailing attraction among the solar Bitcoin types is not just the alleged “free-ness” of solar-generated electricity but also the alleged “greenness” of a zero-carbon hashrate. “Green” electricity would, they think, make Bitcoin not just financially viable to mine but also politically correct.
If they only crunched the numbers.
Maintaining a hashrate in the order of PetaHashes-per-second (PH/s) over any period longer than a second—while competing to solve a block against other miners who are also operating in the PH/s range— requires, collectively across the network, electrical power in the range of megawatts.
An individual miner’s livelihood is inextricably tied to probability: the probability that he/she finds a hash equal to or less than the target, and that his/her solved block of transactions is the one that will be accepted into the Blockchain. At the current time in the half-life of Bitcoin rewards, a miner is rewarded with 6.25 Bitcoins when his/her solved block goes into the Blockchain. (One Bitcoin was worth $8,465 Canadian at 1300 ET on November 13 2017, according to the Canadian Bitcoin Index.)
The entire system is set up to ensure that the probability that any individual miner will solve the puzzle, i.e. find a hash equal to or below the target, for any particular block, is low. This is the entire basis of Bitcoin security.
The probability that an individual miner will solve the block increases with the number of hashes the miner computes. The hashrate is a direct function of the electrical power that makes the ASIC compute hashes: electrical power in, hashes out. The hashrate directly depends on electrical watts.
From this, it follows that if the electrical watts that compute the hashes drop, then so does the hashrate, and with it the miner’s probability that
- his/her equipment will solve the puzzle in the first place, and
- his/her block will become part of the Blockchain.
And in the case of solar power, when the electrical watts stop altogether, as they do every evening when the sun goes down… well, do I even need to say it? Sadly, from my experience with solar enthusiasts, yes I do need to say it. So I will say it right now: when the electrical watts stop altogether, so does the computation of hashes. And so does the solar-powered mining operation. With no electric power to run the computations, the solar-powered miner’s probability of finding a hash equal to or less than the target drops to exactly zero.
With this in mind, I am highly skeptical of the claims made in this Reddit thread and this bitcoin.com article.
The original poster in the Reddit thread claims to run a solar powered greenhouse and Bitcoin mining operation in the desert near San Diego, entirely off grid, using batteries to discharge stored solar power when the sun is down. He claims this is a profitable operation. I simply do not believe that. Here’s what he says:
Each Antminer S9 [the ASIC unit] uses about 1400 watts when it is placed in the environment of our lettuce greenhouse. We place between one and four miners in each greenhouse. We need about 6,000 watts in solar panels to charge up the batteries and simultaneously run one S9 for 24 hours straight. We need 15 of the 24V/200 amp hour batteries for each miner.
So he’s saying that 6 kilowatts of solar power both runs the Antminer S9 during the diurnal insolation and charges fifteen 24-volt lead acid batteries sufficiently for them to collectively discharge power at a rate of 1.4 kilowatts for the—what—14 to 17 hours2 when there is no light hitting the panels meaning they are producing zero power. Fair enough: though he does not ever explain how or even whether he schedules the charged and uncharged batteries so as to cover the panels’ inability to recharge the deployed batteries within the 12-hours-minimum it takes to recharge a lead-acid, it might be conceivable based on those numbers for him to stagger the recharging of his batteries so that each of them receives the roughly 12 to 16 hours required to bring them back to full charge.
But you’ll notice if you have the patience to wade through the interminable thread that he also mentions his batteries also power an air conditioner: his installation, he says, is in the California desert near San Diego, and he needs AC to prevent the Antminers from melting in the daytime heat.
That’s where the BS just becomes obvious. From the numbers he has given, the solar panels cannot run even a 10 kilowatt AC unit; he would need batteries running on stored solar power, and those batteries simply could not hold enough power to run the AC and the Antminer.
Nonetheless, he claims he is profitable. Notice he never provides pictures of his installation, in spite of being asked numerous times to do so.
The DIY-er mentioned in the bitcoin.com article at least provides data and video, from which—when you run the numbers—you can plainly see he is nowhere close to ever making a profit.
The Antminer s3 that he uses consumes around 340 watts of electrical power and computes 441 GH/s, according to this Amazon ad and 355W at 453 GH/s according to this ebay ad.
Let’s stick with the lower wattage/hashrate — 340 and 441 respectively — just for the sake of argument.
The panels in the Youtube video are 220 W each, according to the videographer.
This means that on optimal insolation days (not very frequent in the UK/Northern Europe, from what I hear) he is storing at most ~ 1.4 kWh of solar-generated electrical energy.
The Antminer S3 running at 340 watts eats up those 1.4 kWh in about 4 hours.
Meaning he’s running on power from the grid at least 20 hours on a perfectly sunny day, more than 20 when it’s cloudy.
And we’re supposed to believe this system is “mostly” solar powered?
If he’s in the UK, then his cost of grid electricity is ~12 cents per kWh. Given that on an optimal insolation day he’s getting 1.4 kWh of electrical energy from his panels, and that his Antminer S3 is consuming 8.16 kWh (340 watts x 24 hours divide 1000 to convert Wh to kWh) of electrical energy per day, we can put his cost of electricity at ~10 cents average. I repeat, that’s for an optimal insolation day.
So plug these numbers, especially the $0.1 per kWh, into this Bitcoin mining calculator. To repeat:
- Antminer S3 power consumption: 340 watts.
- Antminer S3 hashrate: 441 GH/s.
- Electricity cost in dollars per kWh: 0.1.
Plug in those numbers… and tell me he’s making a profit.
As you saw, he is not making a profit. (At 1 BTC = $US 5455.35, he attains profitability only when the price he pays for electricity falls below 6.4 cents per kWh — good luck finding that price in any developed country.) Bear in mind my calculation above doesn’t include the capital cost of his solar infrastructure (panels, inverters, batteries, etc.). Those who have purchased panels, inverters, batteries know these things are typically not given away for free, you have to buy them.
Also remember that my calculation assumes a low cost for grid electricity (12 cents per kWh, which works out to 10 cents for 90 percent grid and 10 percent “free” solar). If he is in Germany (~38 cents per kWh) or Denmark (40 cents), his mining operation is going broke much, much faster than it is in the UK.
The entire solar power industry is built on greenwash BS like this.
Meanwhile, as I mentioned above, the Bitcoin network runs on 900 MW of electricity, every minute of every hour of every day. This is for a currency that is barely represented in global transactions.
Those 900 MW are from grids based on 24/7 availability of power. You cannot run a 24/7/365 electrical network using electrical power that is available only 4 to 6 hours per day.
This means that Bitcoin’s only hope of being green and sustainable is for most if not all mining to run on nuclear power.
Which means that for mining—the very activity that makes Bitcoin trustworthy and therefore viable as a currency—to be profitable, it must run on cheap nuclear power.
Which means that electricity must be re-regulated, returned to the rational planning and decisionmaking that existed before the consultants’ free-for-all called deregulation and the rent seekers’ free-for-all called Green Energy.
- As an example of what it takes to find the right nonce: let’s say you are looking for a string whose SHA-256 hash is less than or equal to the following 64-character string (not counting the quotes): “000jc3af42fc31103f1fdc0151fa747ff87349a4714df7cc52ea464e12dcd4e9”. Note that this string begins with three zeros. A suitable candidate could therefore begin with four zeros, followed by a 60-character string that is identical to the 60 characters following the initial “j”, i.e., replace the intial “j” with a zero then follow with the same next 60 characters.
With this problem in mind, what four-character nonce could you concatenate to the end of the string “Hello, world!” that would, when hashed with SHA-256, return such a string? Why, the string “4250” (which puts the input string to “Hello, world!4250” again not counting the quotes).
I only know that because I read this post and loaded my trusty Jupyter Notebook and tried it myself in Python. But if I hadn’t known it beforehand, I would have had to write a function that concatenates a nonce to a string and sees if it matches my requirement and if it doesn’t then increments the nonce by one and tries again till it finds a match. Well, I’m working from an Asus Zenbook UX305 with an Intel® Core™ M 5Y10 Processor. That could have taken a very, very long time.
- The US Energy Information Administration estimated the 2015 annual capacity factor for solar photovoltaic in California to be roughly 28 percent. This Energy Matters article puts that figure lower, but I base my estimates here on 28 percent.
This is a surprise? Anyone who’s tried to debate a few Greens knows that they are almost universally innumerate.
It’s the non-innumerate ones who frustrate me. They know the numbers cold, that’s why they (successfully) lobbied for the extravagant prices. i.e. they know perfectly well that they are peddling an inferior product but are perfectly prepared to profit from it anyway (at other ratepayers’ expense). They slather on the greenwash BS to make it attractive to the innumerate ones—and to governments who pander to the innumerates.
The ones at the top are being paid off. You can’t get them to tell the truth because their paycheck depends on the lie.