In reality, the economics are more complicated.

If an older ASIC has already been fully depreciated, its capital cost is effectively zero. A new miner may consume significantly less electricity per TH, but the efficiency advantage has to recover the cost of the replacement hardware.

I normally think about the upgrade decision using four variables:

1. Electricity savings

Difference in power consumption × electricity price × operating hours.

2. Hashrate increase

New hardware may generate substantially more hashrate using the same rack space and electrical capacity.

3. Expected hardware lifetime

Buying a new machine only makes sense if it remains economically competitive long enough to recover the capital cost.

4. Difficulty and Bitcoin price assumptions

A payback calculation based on today's mining revenue can become meaningless if difficulty rises significantly during the next 12 months.

There is also a strategic consideration: sometimes replacing older hardware is less about immediate ROI and more about extending the period during which a mining operation remains profitable under weaker market conditions.

How do miners here decide when to upgrade?

Do you use a fixed efficiency threshold such as J/TH, a maximum payback period, or simply run older machines until electricity cost makes them unprofitable?

Hashrate alone tells very little about the condition of a machine.

My basic checklist before buying a used ASIC would include:

• individual hashboard performance;
• chip error rate;
• temperature consistency between boards;
• PSU condition;
• fan RPM and bearing noise;
• signs of corrosion or moisture;
• dust contamination;
• repair history;
• previous overclocking;
• firmware currently installed;
• total runtime if reliable logs are available.

Another important point is where the miner was previously operated.

A machine that spent two years in a clean, temperature-controlled facility can be in better condition than a much newer unit that operated in a dusty environment with unstable power.

I would also be cautious with miners where one board produces noticeably lower hashrate than the others. Sometimes it is a simple repair, but sometimes it is the beginning of repeated board failures.

For people who regularly buy second-hand mining equipment:

What are the biggest red flags you look for before buying a used ASIC?

And would you buy remotely based only on logs and video testing, or insist on third-party inspection?

Traditional air cooling is simple and relatively cheap, but once you operate multiple high-power miners, heat management, noise and fan failures become real operational problems.

Hydro and immersion systems solve some of those issues, but they introduce additional complexity.

Air cooling

• lowest initial infrastructure cost;
• simple maintenance;
• easy hardware replacement;
• high noise;
• dust accumulation;
• more difficult heat management.

Hydro cooling

• much lower noise;
• efficient heat removal;
• potentially higher power density;
• additional pumps, plumbing and heat exchangers;
• hardware compatibility becomes important.

Immersion

• excellent thermal stability;
• very low acoustic noise;
• possibility of removing ASIC fans;
• potentially better overclocking conditions;
• higher infrastructure cost;
• more complicated maintenance and fluid management.

For a large mining farm, immersion or hydro can clearly make sense. I am less convinced about the economics for smaller installations.

At what scale do you think advanced cooling becomes economically justified?

Anyone here running immersion or hydro systems with fewer than 20–30 ASICs? I would be interested to hear whether the operational benefits actually justified the investment.

A simple comparison illustrates the problem. Two machines can deliver similar hashrate, but if one consumes several hundred watts more continuously, the difference becomes significant over a full year.

For me, the key metrics when evaluating hardware are:

• efficiency in J/TH;
• actual wall power consumption rather than manufacturer figures alone;
• electricity price;
• expected network difficulty growth;
• machine reliability;
• cooling requirements;
• resale value after 12–24 months.

The interesting question is where the break-even point is.

For example, would you rather buy an older ASIC at half the price but with noticeably worse efficiency, or pay more upfront for a current-generation machine?

I am especially interested in opinions from people running miners at different electricity costs.

At what electricity price do you personally stop considering older-generation ASICs?