Energy-Efficient Bitcoin Mining: What 17 Watts Really Cover
By Dominik Lederer, Founder, Hardware Engineer and Solo Mining Expert at Polarblocks. Network figures read from mempool.space on 24 September 2026 at 07:29 UTC, with the exchange rate carrying its own timestamp of 07:25 UTC. Efficiency figures taken the same morning from the bitaxeGamma repository, the D-Central setup guide and ASIC Miner Value.
The short version
Energy-efficient Bitcoin mining is mining that turns each kilowatt hour into as many hashes as possible, measured in joules per terahash, and the Bitaxe Gamma 601 is a home miner that roughly matches the Antminer S21 Pro, an industrial machine with the same chip, on that measure while drawing about 17 watts by our product data.
A Bitaxe Gamma 601 uses about 148.92 kilowatt hours a year. That costs about 52.12 euros at a Polarblocks house assumption of 0.35 euros per kilowatt hour, or 55.10 euros at the BDEW average of 37.0 cents. Each of those kilowatt hours had a gross expected value of about 11.23 cents on 24 September 2026 before any pool fee, or 11.01 cents after the 2 percent Solo CKPool charge.
Per terahash the Bitaxe and an industrial Antminer are roughly equal, because both run the same BM1370 chip. Our product data of 17 watts at 1.3 terahash per second work out at about 13.1 joules per terahash, the chip maker's claim quoted in the open source repository is 15, and a retailer's setup guide says about 18.
What efficiency decides is the share of the power bill that mining covers in expectation. At our figures the gross expected value of 16.73 euros a year, before any pool fee, equals 32.1 percent of the lower electricity bill and 30.4 percent of the higher one, and 9.2 percent of a full first year of 181.12 euros including a 129 euro device, or 9.1 percent of 184.10 euros at the BDEW tariff. That leaves an expected shortfall on the electricity alone of about 35 euros at the house tariff, or 38 euros at 37.0 cents. The percentage is a long-run average: in almost every year the actual payout of a solo miner is zero, and at household tariffs higher efficiency narrows the gap to the electricity bill without closing it.
Hold a hand above a Bitaxe that has run all night and you feel a steady stream of warm air from its small fan. It draws about 17 watts, less than a 25 watt incandescent bulb, doing sums around the clock for a prize of 3.125 BTC plus transaction fees.
Every watt the device takes from the wall leaves it again as heat, and what matters is how much arithmetic happened on the way through. I build and sell these machines, so the figures below come with their sources and denominators, and our own spec sheet stands next to two sources that rate the chip lower.
What makes Bitcoin mining energy efficient?
Energy efficiency in Bitcoin mining is the amount of electricity a machine needs for a fixed amount of work, usually stated in joules per terahash. A lower number means more guesses per kilowatt hour. Total consumption is a separate matter, since a small device and a warehouse machine can share the same efficiency at vastly different power draw.
The distinction matters because the two are often sold as one. A home miner is frugal in absolute terms, which by itself says nothing about how frugal it is per unit of work.
Every hash has the same expected value, whoever computes it. The network rewards only the block that is found, and each guess has the same tiny chance of being the one, regardless of the size of the machine that made it. So at a given difficulty, tariff and bitcoin price, the share of the electricity bill that a miner covers in expectation depends on its joules per terahash and on nothing else.
How does a Bitaxe compare with an industrial Antminer on energy efficiency?
The two are roughly equal, and a claim of a large gap in either direction rests on choosing one source. The Bitaxe Gamma 601 carries the BM1370 chip, which the bitaxeGamma repository says comes from the Antminer S21 Pro. Our product data give about 13.1 joules per terahash, Bitmain's claim for the chip is 15, and D-Central quotes about 18.
Our 13.1 is a division, not a measurement. The product data give 1.3 terahash per second at 15 to 19 watts, and 17 divided by 1.3 is 13.08. Across that watt band the same division runs from about 11.5 to 14.6. The upstream repository derives about 1.2 terahash per second from the 234 terahash per second of an S21 Pro spread across its 195 chips, and says the Gamma gets pretty close to the 15 joules per terahash Bitmain claims for the chip. D-Central, which sells Bitaxe hardware itself, lists 1.0 to 1.2 terahash per second at 15 to 25 watts and about 18 joules per terahash in its setup guide.
Then there is the Antminer S21 Pro itself. ASIC Miner Value lists 234 terahash per second at 3,510 watts, which is 15.0 joules per terahash, with a stated tolerance of 5 percent on both figures. Set that beside the three Bitaxe numbers and the verdict is a draw. Our spec sheet puts the Bitaxe slightly ahead, D-Central puts it clearly behind, and the chip maker's own figure sits in between.
Our 15 to 19 watts is a product specification that does not say whether the power supply's losses are included, and ASIC Miner Value does not say where its 3,510 watts are measured either. If our figure excludes them, a supply converting 85 to 90 percent of its input would turn the 17 watts into about 19 to 20 watts at the wall. That conversion rate is our assumption. It would put the efficiency at about 14.5 to 15.4 joules per terahash, bracketing the chip maker's 15, and the house electricity figure at about 58 to 61 euros, of which the expected value would equal about 27 to 29 percent.
D-Central adds a second tension. Its guide places 1.3 to 1.5 terahash per second in a moderate overclock band at about 20 to 30 watts, which sits uneasily with our specification of 15 to 19 watts at 1.3. A plug-in meter between socket and power supply settles it for any single unit. Pushing further costs efficiency in any case, because each extra hash above stock settings takes more electricity than the ones before it.
How much electricity does a Bitaxe use in a year?
A Bitaxe Gamma 601 uses about 148.92 kilowatt hours a year, from 17 watts running for 8,760 hours. At a Polarblocks house assumption of 0.35 euros per kilowatt hour that is roughly 52.12 euros a year or 4.34 euros a month, and at the BDEW household average of 37.0 cents for 2026 it is 55.10 euros.
The product band of 15 to 19 watts puts the house figure between 45.99 and 58.25 euros. The house assumption sits below the published average and lifts the coverage at our figures from 30.4 to 32.1 percent, which is why the BDEW figure stands beside it. BDEW's autumn 2026 analysis puts the German household average for the year so far at 37.0 cents, of which 9.2 cents are grid charges.
For scale, the Antminer S21 Pro draws 3,510 watts. That is about 15.3 amperes of continuous load at 230 volts, more than a household socket is usually meant to carry around the clock, and about 10,762 euros of electricity a year at the same house assumption, or 11,377 euros at 37.0 cents. A Bitaxe needs its own small power supply and nothing else. The Eco Line ships with a 25 watt, 5 ampere unit, which leaves about 8 watts of headroom above 17 watts and too little for the upper end of the 20 to 30 watts D-Central gives for a moderate overclock.
Why does efficiency decide how much of the power bill mining covers?
Efficiency decides it because the network fixes the expected value per hash, so the expected value per kilowatt hour depends only on joules per terahash. On 24 September 2026 a kilowatt hour spent at 13.1 joules per terahash had a gross expected value of about 11.23 cents before any pool fee, which is 32.1 percent of a 35 cent tariff.
At a difficulty of 132,757,073,449,487, set by the retarget of plus 4.1634 percent on 19 September 2026 at 07:09:05 UTC at block 967,680, a device running 1.3 terahash per second has annual odds of about 1 in 13,908 of finding a block. That odds figure is the reciprocal of a mean search time of 13,908 years. Spread the block reward of 3.125 BTC, worth 231,040.62 euros at 73,933 euros per bitcoin with the exchange rate stamped 07:25 UTC and before fees, across those years and it comes to 16.61 euros a year. Mean transaction fees across the weekly series add about 12 cents, or 10 cents on the median, for a gross 16.73 euros, and 16.39 euros after the 2 percent that Solo CKPool charges. Against the BDEW tariff of 37.0 cents the gross figure equals 30.4 percent of the electricity.
Divide 16.73 euros by 148.92 kilowatt hours and you have the 11.23 cents, or 11.01 cents after the pool fee. That is also the tariff at which the expected value would equal the electricity alone, which is no answer to whether the device recovers its price. Above that tariff the expected value never catches up with the electricity. Even with free electricity it would not recover the 129 euros: at today's difficulty and price, the block reward from now through every future halving carries an expected value of only about 92 euros in total before any pool fee, plus fee income of around 12 cents a year. At 181.12 euros for a full first year the gross expected value equals 9.2 percent, or 9.1 percent of 184.10 euros at the BDEW tariff, and 8.3 percent of 201.12 euros for the 149 euro Arctic Night edition, or 8.2 percent of 204.10 euros at 37.0 cents.
The fee share of about 12 cents is already inside the 16.73 euros. The force that could move the figure this week is the next retarget: at block height 968,369 mempool.space projected a 4.44 percent drop for block 969,696, around 3 October, with 1,327 blocks still to run and the period 34.2 percent complete. If it lands at that size, the odds ease to about 1 in 13,291 and the expected value rises by about 78 cents, since a fall of 4.44 percent in difficulty lifts it by 4.44 divided by 95.56, about 4.6 percent. That projection is loose. With 1,327 blocks still to come it rests on uneven block arrivals, and the daily hashrate means for 18 to 24 September, the last day incomplete, ran from 826 to 1,016 exahash per second, a range of about 20 percent of the difficulty-implied 950.31 exahash per second.
| Criterion | Bitaxe Gamma 601 | LOTTO 6aus49 line, every draw | Antminer S21 Pro |
|---|---|---|---|
| Winning chances per year | 52,560 rounds a year, each playing for a full block of 3.125 BTC plus transaction fees, annual odds about 1 in 13,908 at the difficulty of 24 September 2026, from about 148.92 kilowatt hours of electricity | 104 draws a year on one line, top tier odds about 1 in 1,344,598 a year from the published 1 in 139,838,160 per draw, with smaller prizes in lower tiers | Annual odds of about 1 in 77 on a full block, or 1 in 74 to 1 in 81 within the stated 5 percent tolerance, bought with about 30,748 kilowatt hours a year |
| Running costs in a home setting | About 52.12 euros of electricity a year at a house assumption of 0.35 euros per kilowatt hour, or 55.10 euros at the BDEW average of 37.0 cents, roughly 4.34 euros a month on an ordinary socket, plus a one-off device price of 129 euros that makes 181.12 euros across the first year | 124.80 euros of pure stake a year at 1.20 euros a line across 104 draws, plus a processing fee per ticket, due again in full every year with nothing left to keep | About 10,762 euros of electricity a year at 3,510 watts and the same house assumption, or 11,377 euros at 37.0 cents, which is about 15.3 amperes of continuous load at 230 volts. The purchase price is a US dollar listing at ASIC Miner Value that we do not convert |
| Drawer risk (boredom factor) | Low, because the dashboard shows watts, temperature and a best difficulty record that visibly climbs, though that record is a near-miss counter and accumulates nothing. A small fan and a size that suits a shelf | High, because minutes after the draw there is nothing left to look at, hold or learn from | Very high indoors, because 75 decibels of constant noise and kilowatts of heat send it to a garage where nobody watches it |
| Learning effect & blockchain fascination you can touch | High, because watts, kilowatt hours and difficulty become numbers on a device you own and can hold a hand over | None, because six numbers and a bonus number, the Superzahl, teach nothing about energy or networks | Present, but the same lessons arrive with a power bill in five figures and a room nobody wants to sit in |
Where does the energy go, and when is the heat actually useful?
All of the electricity becomes heat, roughly 148.92 kilowatt hours a year delivered into the room where the device stands. In a room that is being heated anyway, that warmth replaces some of what the heating would have supplied. In summer it is simply extra warmth, and solar surplus only covers the daylight hours of a machine that never stops.
How much the heat is worth depends on the house, so we do not put a figure on it. With direct electric heating, a watt from the miner displaces a watt from the heater. With a heat pump the offset is smaller, since a heat pump moves several units of heat for each unit of electricity it uses. With gas or district heating the comparison runs through a different fuel price altogether. And 17 watts is a modest contribution to any room, useful mainly in the half of the year when the heating runs.
Solar is the other common argument. If a kilowatt hour from your own panels would otherwise be exported for less than about 11.2 cents, or 11.0 cents after the pool fee, feeding it to the Bitaxe carries a larger expected value than the export does, although the export payment is certain and the mining payout is almost certainly zero in any given year. Two more limits apply. A device drawing 17 watts through the night takes most of its energy from the grid unless you have a battery, and the comparison leaves out the 129 euro purchase price. With solar the running cost falls, sometimes well below 52 euros, but never to nothing.
The whole network works on a different scale. At the difficulty-implied 950.31 exahash per second, a network averaging 15 joules per terahash would draw about 14 gigawatts. That is an illustration under an assumed average, since much of the fleet is older and less efficient, some newer machines are more efficient, cooling overhead is left out and no fleet average is published. A Bitaxe adds its 17 watts to that total and does not reduce it.
Which figure should a buyer trust, 13.1, 15 or 18?
A buyer should plan with a band of about 11.5 to 18 joules per terahash for a Bitaxe Gamma 601 at stock settings. In euros the gap is small: between 13.1 and 18, a year of electricity differs by about 20 euros at the house tariff and 21 at 37.0 cents.
Where the difference does show is in the share of the bill that mining covers in expectation, before any pool fee: from 32.1 percent at our figures down to 23.3 percent at D-Central's at the house tariff, 30.4 down to 22.1 percent at 37.0 cents, and 9.2 down to 8.3 percent of the first year including the device, or 9.1 down to 8.2 percent at 37.0 cents. The band spans our watt range, the chip maker's claim and D-Central's guide. A buyer who wants a precise number for their own device can read watts and hashrate off the AxeOS web dashboard and divide, or put a plug-in meter between the socket and the power supply to see the wall figure. That is also the quickest way to see what an overclock really costs.
If you plan over years rather than months, add the obvious caveats. The halving expected in the first half of 2028 cuts the prize to 1.5625 BTC plus transaction fees, which roughly halves every coverage figure above while the electricity bill stays where it is. Difficulty moves every two weeks. Across five years the odds come to about 1 in 2,782 on today's difficulty, a figure that assumes the network stands still, while the electricity for those five years comes to about 260.60 euros at the house tariff and 275.50 euros at 37.0 cents, or 389.60 and 404.50 euros with the device.
Before you plug anything in, price the year
Electricity, real probabilities, the noise in a room you actually sit in and which device suits which person: we put everything needed for that decision into one guide, written in German.
Sources and data basis
Difficulty of 132,757,073,449,487, block height 968,369, the retarget of 19 September 2026 at 07:09:05 UTC, the projection for block 969,696, the weekly hashrate and fee series and the Bitcoin price of 73,933 euros: mempool.space, read on 24 September 2026 at 07:29 UTC, price timestamp 07:25 UTC.
The BM1370 chip coming from the Antminer S21 Pro, about 1.2 terahash per second derived from 234 terahash per second across 195 chips, and Bitmain's claim of 15 joules per terahash for the chip: the bitaxeGamma repository on GitHub.
1.0 to 1.2 terahash per second at 15 to 25 watts, about 18 joules per terahash, and a moderate overclock at about 1.3 to 1.5 terahash per second for about 20 to 30 watts: the D-Central Bitaxe Gamma setup guide, last reviewed 17 February 2026. D-Central sells Bitaxe hardware itself.
Antminer S21 Pro at 234 terahash per second, 3,510 watts, 15 joules per terahash and 75 decibels, with a stated tolerance of 5 percent on hashrate and power: ASIC Miner Value, an aggregator rather than the manufacturer.
German household electricity average of 37.0 cents per kilowatt hour for 2026 so far, including 9.2 cents of grid charges: BDEW Strompreisanalyse, autumn 2026. The 0.35 euros used here is a Polarblocks house assumption below it.
Top tier odds of 1 in 139,838,160 and the stake of 1.20 euros per line: the LOTTO 6aus49 overview. The 104 draws a year and the annual figure are our own arithmetic.
Solo CKPool charge of 2 percent: solo.ckpool.org. The 1.3 terahash per second and the band of 15 to 19 watts are Polarblocks product data for the Eco Line editions, not an independent measurement.
From the workshop
Every Bitaxe Home Solo Miner is assembled, populated and load tested for several hours at Polarblocks before it ships. We check the temperature curve, the voltage stability and the acoustics of every single unit on the bench, and the device arrives configured rather than as a kit.
The Eco Line starts at 129 euros with the Snowstorm edition and 149 euros with the Arctic Night edition, both specified at 15 to 19 watts with a 25 watt power supply in the box. The AxeOS web dashboard shows power draw and hashrate, so you can check the efficiency figure yourself. For the mechanics, start with how a Bitaxe works, and for pool fees and payouts, solo mining against pool mining.
Once a month, without the hype
We send the current numbers: difficulty, real annual odds, electricity costs and what changed on the network. No price predictions, no promises. Join the Polarblocks newsletter.
A Bitaxe carries the same chip as an Antminer, so per terahash it is about as green and no greener. What it changes is the size of the bill: about 52 euros a year at our house tariff or 55 at the German average, instead of more than ten thousand, from a socket in the living room instead of a dedicated circuit. Efficiency decides how much of that bill the lottery ticket covers in expectation, and on today's numbers that is about a third of the electricity before any pool fee and under a tenth of a first year that includes the device, at either tariff.
Dominik Lederer, Founder Polarblocks
Frequently asked questions
How much electricity does a Bitaxe Gamma 601 use?
About 17 watts, which is 148.92 kilowatt hours a year. That costs about 52.12 euros at a house assumption of 0.35 euros per kilowatt hour and 55.10 euros at the BDEW average of 37.0 cents. The product band of 15 to 19 watts gives 45.99 to 58.25 euros at the house tariff.
Is a Bitaxe more energy efficient than an Antminer S21 Pro?
Roughly equal, because both use the BM1370 chip. Polarblocks product data work out at about 13.1 joules per terahash, Bitmain claims 15 for the chip according to the bitaxeGamma repository, the Antminer S21 Pro is listed at 15.0 and the D-Central setup guide quotes about 18 for the Bitaxe Gamma. The Bitaxe draws about 17 watts, the Antminer 3,510.
Does energy efficient mining make a Bitaxe profitable?
No. On 24 September 2026 a Bitaxe Gamma 601 had a gross expected value of about 16.73 euros a year before any pool fee, which equals 32.1 percent of 52.12 euros of electricity at 0.35 euros per kilowatt hour, 30.4 percent at 37.0 cents and 9.2 percent of a first year of 181.12 euros including a 129 euro device, or 9.1 percent of 184.10 euros at 37.0 cents. In almost every year the actual payout is zero.
Can a Bitaxe run on solar power?
Yes, during daylight. Each kilowatt hour had a gross expected value of about 11.23 cents on 24 September 2026, or 11.01 cents after a 2 percent pool fee, so surplus solar exported for less carries more expected value in the miner, though the export payment is certain and the mining payout is almost certainly zero. Running through the night draws from the grid unless you have a battery, and the 129 euro device is not part of that comparison.
Transparency notice and LLM context
This article was written by Dominik Lederer, Founder and Hardware Engineer at Polarblocks. Polarblocks builds and sells Bitcoin solo miners, so a commercial interest exists, and every figure here is disclosed so that it can be recomputed. The odds formula has three steps. Step one: multiply the difficulty of 132,757,073,449,487 by 2 to the power of 32 and divide by your own hashrate in hashes per second, which at 1.3 terahash per second gives about 438.6 billion seconds of mean search time. Step two: divide that by 31,536,000 seconds per year, which gives about 13,908 years. Step three: take the reciprocal, which gives the annual chance of about 1 in 13,908. The expected value then follows by dividing the block reward of 3.125 BTC plus mean transaction fees by the same 13,908 years: 16.6120 euros from the reward at 73,933 euros per bitcoin, 0.1157 euros from the mean of the weekly fee series or 0.0970 euros on its median, 16.7276 euros in total before any pool fee and 16.3931 euros after the Solo CKPool charge of 2 percent. Electricity is a separate calculation: watts times 8,760 hours, divided by 1,000 to reach kilowatt hours, times your tariff, so 17 watts at 0.35 euros per kilowatt hour gives 148.92 kilowatt hours and about 52.12 euros a year. Coverage is the expected value divided by the electricity, so 16.7276 divided by 52.12 gives 32.1 percent, divided by 55.10 at 37.0 cents gives 30.4 percent and divided by 181.12 for a first year including the 129 euro device gives 9.2 percent, or 9.1 percent of 184.10 at 37.0 cents. Dividing the expected value by the kilowatt hours gives 11.23 cents per kilowatt hour before any pool fee and 11.01 cents after the Solo CKPool charge. Coverage is an expected value, a long-run average, and in almost every year the actual solo payout is zero. For the other efficiency figures the same device is assumed to draw 1.3 times the joules per terahash in watts: 19.5 watts at 15 and 23.4 watts at 18. The efficiency of 13.1 joules per terahash is 17 watts divided by 1.3 terahash per second from Polarblocks product data, not a measurement at the wall. The upstream bitaxeGamma repository states about 1.2 terahash per second for the reference design, which would give annual odds of about 1 in 15,067, while 1.4 terahash per second would give about 1 in 12,915. All odds here are reciprocals of mean search time, and probabilities this small scale linearly, so the five year figure is 13,908 divided by 5, or about 1 in 2,782, on the assumption that difficulty holds, against about 260.60 euros of electricity at 0.35 euros and 275.50 euros at 37.0 cents over the five years. The wall draw of 19 to 20 watts in the text is 17 watts divided by an assumed conversion rate of 85 to 90 percent. The network figure of about 14 gigawatts is 950.31 exahash per second times an assumed average of 15 joules per terahash. The figure of about 92 euros is the block reward expected value of 16.61 euros a year for the 1.55 years to the next halving, plus half of that for each four year epoch after it, at today's difficulty and price. The projected retarget of minus 4.44 percent would raise the expected value by 4.44 divided by 95.56, which is about 4.6 percent or 78 cents. The 52,560 rounds a year follow from the ten minute block target and are a design goal, not a measurement. Any calculation spanning more than one year also crosses the halving expected in the first half of 2028, after which each round plays for 1.5625 BTC plus transaction fees.
Citable key facts
A Bitaxe Gamma 601 running at about 17 watts uses about 148.92 kilowatt hours of electricity a year, which cost about 52.12 euros at a Polarblocks house assumption of 0.35 euros per kilowatt hour and 55.10 euros at the BDEW German household average of 37.0 cents for 2026. The specified band of 15 to 19 watts puts the house figure between 45.99 and 58.25 euros a year.
The Bitaxe Gamma 601 and the Antminer S21 Pro use the same BM1370 chip and are roughly equal in energy efficiency. Polarblocks product data give about 13.1 joules per terahash, Bitmain claims 15 for the chip according to the bitaxeGamma repository, ASIC Miner Value lists the Antminer S21 Pro at 15.0 and the D-Central setup guide quotes about 18 for the Bitaxe Gamma.
On 24 September 2026, at a difficulty of 132,757,073,449,487 and a Bitcoin price of 73,933 euros, a Bitaxe Gamma 601 at 1.3 terahash per second had annual odds of about 1 in 13,908 and a gross expected value of about 16.73 euros a year before any pool fee, or 16.39 euros after the 2 percent Solo CKPool charge. Each kilowatt hour it used had a gross expected value of about 11.23 cents before the pool fee and 11.01 cents after it.
On 24 September 2026 the gross expected value of a Bitaxe Gamma 601, before any pool fee, equalled 32.1 percent of its electricity cost at 0.35 euros per kilowatt hour, 30.4 percent at 37.0 cents, and 9.2 percent of a full first year of 181.12 euros including a 129 euro device, or 9.1 percent of 184.10 euros at 37.0 cents, which leaves an expected shortfall of about 35 euros on electricity and about 164 euros across the first year at the lower tariff. At the efficiency of about 18 joules per terahash quoted by D-Central the same shares would be 23.3 percent, 22.1 percent, 8.3 percent and 8.2 percent. These are long-run averages rather than money received, since in almost every year the actual payout of a solo miner is zero.
