Home / Blog / Bitmain Antminer S17
Field report · Sep 5, 2026 · 7 min read
The Antminer S17 mined the whole bull run in Argentina, then crossed the Atlantic once
Three and a half years on a high-uptime site, straight through the best market the network has ever had, and then one crossing to cheap power when a full year of figures said it belonged there. It mined until November 2025, and the coins are all still here.
Bitmain Antminer S17 (56 TH/s)
Seven nanometres, and a new idea of efficient
When Bitmain introduced the S17 range in April 2019 it was the first of its machines on 7-nanometre chips, and it roughly halved the power per terahash of the S9 generation. The base S17 does 56 TH/s at 45 J/TH — in 2019, a genuinely efficient machine.
We bought ours in October 2019 for $1,248, and it landed in Argentina at $2,485 once freight, duty, infrastructure and setup were in.
It had a rough reputation in its first year; the forums are full of temperature-sensor threads and board complaints. That is the kind of problem a site with a repair bench absorbs as routine work, and it is not what this machine’s story turned out to be about.
One crossing, and the three years it waited
It spent its first 1,126 days in Argentina — three and a half years, including every month of the 2021 bull run. That is the part people expect to be wrong. Nigeria’s power is a third cheaper, so surely the machine should have gone there immediately?
It should not, and the bull run is why. When hashprice is high, every hour a machine is hashing is worth a great deal, and Argentina’s better-than-98% uptime mines enough extra coins to beat a cheaper power bill outright. Through 2021 the expensive site was comfortably the better site, and a machine sitting in Nigeria would have earned less.
What changed it was not a bad quarter but a bad era. As hashprice fell away from the 2021 peak and stayed down, the trailing year stopped being close: cheap power was winning, and winning by a distance that no rally reversed. On 19 March 2023 the machine was crated up and sent to Nigeria, where it spent the 929 days that were left to it. One crossing, made once the answer had stopped changing.
While a terahash still earned well, uptime paid more than cheap power. The machine did not move until that had stopped being true for a full year.
A long last stretch
The first time it stood still was before the crossing, not after it. On 14 November 2022, with hashprice at its lowest since the bull run and the machine still in Argentina, nothing we hosted covered its power for a month — so it was switched off rather than run at a loss. It came back on that winter, ran another seven weeks in Argentina, and then made the crossing.
In Nigeria it mined for another two and a half years, standing down whenever the arithmetic said to and coming back when it turned, and was switched off for good on 17 November 2025. Across its life it stood idle for 427 days, which is the reason it never mined a single day at a loss.
The box was worth more at the top than we paid for it
The 2021 bull run did not only lift bitcoin. It lifted the price of the machines that mine it. In March 2021, eighteen months after we bought ours, an S17 was fetching $5,400 — 4.3× what we paid for it.
An owner who sold there would have walked away with 0.3324 BTC already in the wallet — worth $26,970 today — plus $5,400 for the box: $32,370, 13.0× the landed cost, for a year and a half of hashing. That is the part of mining people outside it rarely see: at the top of a cycle, the hardware itself can return a multiple of its price on top of everything it mined.
We kept ours running instead, and it went on mining for years — which is why its coins alone are now worth $46,981. Selling at the top and keeping on mining were both very good outcomes. The only bad one would have been not owning it.
Mine it for eighteen months, sell the box at the top, keep the coins. Or keep mining. Both were extraordinary.
The tally
Across 2,107 days of mining it kept 0.5790 BTC for $6,124 of electricity — about $10,575 a coin. Those coins are worth $46,981 today.
Against a $2,485 landed cost, that is $40,858 of cashflow — 16.4× its hardware. 0.2083 BTC of it, 36% of everything it made, came in its first seven months, before the 2020 halving.
Why the timing mattered more than the move
A machine that had gone straight to the cheapest power would have done well. This one did better, because for three and a half years the more expensive site was genuinely the better one, and the machine was not moved out of it on a hunch. The cheapest power is not automatically the best site. The best site is whichever turns a day of hashing into the most value — and early in a machine’s life, with hashprice high, that is usually the site that almost never goes down.
That is the same rule our fleet runs on today. An S21 or an S19 XP Hyd crosses once, when a year of numbers says so rather than when something breaks — and the S17 is the clearest picture we have of what that patience is worth over a whole working life.
Sources and further reading
- Bitmain introduces the Antminer S17 Pro, Antminer S17 and Antminer T17 — Bitcointalk, April 2019
- Antminer T17/S17 temperature sensor discussion — Bitcointalk, 2020
Which machines this covers
Disclosure
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.