Used Antminers: what drives the price and what to check

Used mining hardware is a real market with real logic behind it. Machines come out of operators who are upgrading, exiting or consolidating, and they arrive in batches with a history that is usually only partly documented. Understanding what sets the price is most of what you need to buy well.

What actually drives the price

• Efficiency class, first and by a long way. A used machine is priced against what it will earn, and that is set by joules per terahash against the current market. Age matters mainly because it correlates with efficiency.
• Where hashprice sits. Used prices move faster than new prices in both directions, because the marginal machine is the first to become unviable and the first to come back.
• The calendar. Prices soften before a halving as everyone runs the same arithmetic on the same models. See what the halving does to a fleet.
• Batch size and consistency. Two hundred identical units with one test record are worth more per unit than two hundred mixed machines from four sources.
• What the seller will warrant. A batch with a written dead on arrival window prices above one sold as seen, and it should.

The vocabulary problem

Refurbished, reconditioned, tested, grade A, like new. None of these terms is regulated and none of them means anything without a definition attached. In the market, refurbished usually implies a rebuild with component replacement and a burn in report, and a great deal of stock is sold as refurbished having received neither.

We call our stock used, deliberately, for that reason. It means previously operated hardware, function tested before dispatch, with the condition written on the quote and a dead on arrival warranty of ten days from delivery. If a seller uses a stronger word, ask them to write down exactly what was done and what they are standing behind. The full argument is in new or used.

What to check before you pay

• Photographs of the actual units with serial numbers, taken on request rather than pulled from a folder.
• The test record. What was tested, how, and what the results were. Per unit, not a blanket statement about the batch.
• Hashboard status. A machine running on two of three boards still reports as online. Ask for figures per board.
• Power supply condition and voltage rating. The PSU is the component most likely to have been stressed, and it must match your site. See ASIC power supplies.
• Firmware. Whether the machines are on stock or third party firmware, and whether they were overclocked. Both affect remaining life.
• Where they ran. A clean hydro site and a dusty container produce very different machines of the same age.

The claim window is the whole protection

With used stock the warranty is short by design. That makes the first days after delivery the only period in which a fault is someone else's problem, so the entire consignment should be powered on and checked immediately, including the units you do not need yet and the spares. Operators who energise in stages routinely discover a fault in week three and have no claim. The sequence is in the first 72 hours.

How much life is left

The question is not how old the machine is but where its breakeven sits against your power price, and how far the network has to move before it stops paying. On cheap power a used machine can have years left. On grid rates it can be finished inside a year. Work it out before you buy, using the breakeven power price and how long an ASIC miner lasts.