Home / Blog / Bitmain Antminer S19K Pro
Field report · Sep 20, 2026 · 8 min read
Bought the week it launched, racked in Argentina at 6¢, run through a halving, moved to Nigeria when it stopped clearing its power. Not one sat sold. 0.1151 BTC kept, $4,209 of electricity paid, and about $1,888 up on everything that went in.
Bitmain Antminer S19K Pro (120 TH/s)
Bitcoin was around $26,000. The halving was a date on a calendar that nobody was pricing in yet. Bitmain had just put the Antminer S19K Pro on the market at about $1,820 for the 120 TH version, and after two years of $3,000 machines it felt like the door had been opened. Fifteen dollars a terahash. You could start mining for the price of a used car engine.
Landed at our Argentina site — machine, freight, import duty, customs, rack space and setup — the all-in was $3,257. Argentina charges import duty and a broker fee, and together they add almost as much again as the freight. That is the number that matters, and it is the number most buyers never write down. Everything that follows is measured against it.
The site itself is the unglamorous half of the story. Six cents a kilowatt-hour, and over the past twenty-four months it has run above 99% uptime. Uptime is the variable nobody markets and everybody pays for: a machine that is off is a machine paying rent on capital while producing nothing.

From launch to the halving in April 2024 the machine ran 263 days and put 0.0641 BTC in the wallet, clearing about $5.72 a day after electricity. At that rate the arithmetic looked wonderful — a little over a year to pay for itself, and then free money.
That arithmetic was a lie, and the reason it was a lie was printed in the protocol years in advance. On 20 April 2024 the block subsidy halved, and every machine on earth started producing half as many coins for the same electricity.
The 884 days since have produced 0.0587 BTC — slightly less than the first 263 days produced — at about $1.32 a day after power. Same machine, same site, same uptime. The first 23% of its working life delivered more than half of everything it has ever made. If you take one thing from this piece, take that sentence and apply it to whatever you are being sold today.
It produced more than half of every coin it would ever produce in the first 23% of its life.
What follows a halving is not a crash, it is a grind. Difficulty kept climbing because better machines kept arriving. Our machine kept doing exactly what it was built to do, at a slowly shrinking share of a slowly growing network. There was no drama. There was just a daily number that got smaller.
Meanwhile the hardware was in free fall. The same box that cost $1,820 in August 2023 was worth around $900 by mid-2025 and under $400 by the time of writing. Anyone marking their mining position to market was watching a disaster. Anyone counting coins was fine.
That gap between the two ways of looking at the same machine is the real subject of this article. Judged as an asset it lost 79%. Judged as a converter that turns cheap electricity into bitcoin, it just kept converting.
Across 1,131 days it mined and kept 0.1151 BTC. Not one sat was sold — the power bill was paid out of pocket, which is the only way to run this if you actually want the bitcoin. Electricity came to about $4,209. The hardware was $3,257 landed. Those coins are worth about $9,354 today.
Coins, less electricity, less the hardware: up about $1,888 on $7,466 of total outlay — a 25% return, with the machine still running and the box still worth a few hundred on top. That is not spectacular. It is a real return on a machine everybody wrote off, earned over three years by a box that cost less than a used car engine.
Averaged over the whole run, each coin cost about $36,600 in electricity — against a market price north of $80,000. That is the number that makes this work: every sat was manufactured at well under half what it trades for. It is a lifetime average and not what the machine does today; at today’s difficulty a coin costs it far more, which is why it has moved sites and why it is in its last chapter.
At 6¢ the S19K Pro is now close to the edge. In a weak month it clears very little, and there are months where it is not obviously worth the rack space that a newer machine could use instead. In Argentina its working life is more or less over.
But cost per coin is a function of the power rate, and we run a site in Nigeria at 3.9¢. Moving the same machine there takes its electricity cost per bitcoin from roughly $38,500 to roughly $25,000 — the machine did not change, the bill did. Nigeria runs a lower uptime than Argentina, so it mines fewer coins per year, but it mines them substantially cheaper, and for a machine in its last chapter that is the trade you want.
This is the part of the lifecycle people miss. Machines do not die when they stop being competitive at your rate; they migrate to cheaper power and keep producing until they stop producing. Our older fleet finishes its working life in Nigeria for exactly this reason. Sooner or later every one of them reaches a floor where no rate is cheap enough — that is the trip they are all on — but it is usually years later than the resale price suggests.
An old machine is not finished. It is finished at that price of electricity.

Buy the more efficient machine. Every time. The S19 XP cost more in 2023 and is still mining profitably at rates where this one is not. Efficiency is not a spec-sheet nicety, it is the number that decides how many halvings you survive.
Write down the landed cost. The $1,820 machine was a $3,257 commitment once freight, import duty, customs and setup were in — 79% more than the sticker — and the payback clock runs on the larger number. It is the difference between this machine having covered itself comfortably and being at 82%.
Do not annualise a pre-halving day rate. If you had taken that $5.81 a day and multiplied it by 365 you would have expected $2,088 in year one alone. It took three years to produce $2,670 — a good result, and less than half the pace the first eight months implied.
And do not buy the dip on hardware. The same machine bought six months later at $1,513 is down 12%, and a year later at $1,279 is down 31%. Cheaper box, worse outcome, every time — because the coins you miss by waiting are the pre-halving ones, and those are the only coins that were ever going to carry a 23 J/TH machine. Timing beat price, and neither beat the power rate.
For what we think of the hardware itself — the failure rates, the board revisions, what breaks — read the Bitmain Antminer S19K Pro review. For why we think about any of these machines the way we do, machines are the roof, not the building.
Bitmain Antminer S19K Pro review — Mediocre hardware that made money anyway — if you bought it on day one and kept the coins. The launch-day unit is up 25% on everything that went into it. The same machine bought six months later, for $300 less, is down 12%.
Firsthand Bitcoin sells and hosts mining hardware, including this machine. No manufacturer, distributor or affiliate programme paid for or reviewed this page and we take no commission on the links above. Historical figures are computed from daily bitcoin price and network hashprice, each day valued at its own prices. Nothing here is investment advice.