ASIC payback: how to calculate ROI without mistakes

ASIC payback is the price you paid divided by what the machine clears per day after electricity. The number every calculator hands you assumes difficulty stays frozen. It never does. After the 19.09.2026 retarget, difficulty rose 4.16% to 132.76 T, and spot hashprice sat at $40.93 per PH per day (Hashrate Index, 21.09.2026). Your real payback period is longer than any single-snapshot projection, and the gap widens with every retarget. This page walks through the formula, the corrections most people skip, and a worked comparison across three current models at three electricity rates.

POOL BTC is an independent comparison site. We take no hashrate, pay no rewards, and sell no hardware.

The formula everything else hangs off

Payback in days = X / (D minus E), where

• X is the full cost of entry: the machine, shipping, duties, a PSU if it ships separately, and the wiring to carry the load.
• D is daily revenue after the pool fee.
• E is daily electricity cost.

The rest of this article is corrections to D and E. There are five: difficulty growth, downtime and rejected shares, the gap between datasheet and real hardware, withdrawal fees plus exchange spread, and the payout scheme you pick. All five push the result the same direction, except payout scheme variance on PPLNS, which swings both ways.

How to calculate ASIC payback, step by step

1. Fix X. Not the sticker price on the manufacturer's site. The amount that will actually leave your account: machine, shipping, duties, electrical work. If the unit goes to a hosting site, the upfront deposit belongs here too.
2. Take rated TH/s and watts. Models with their hashrate and power draw are listed in the ASIC miner ranking. Efficiency in J/TH is watts divided by TH/s.
3. Work out D. BTC revenue for your hashrate at current difficulty, minus the pool fee. You do not need to do this by hand: the profitability calculator does it and shows FPPS against PPLNS on the same inputs.
4. Work out E. Watts times 24 divided by 1000 gives kWh per day. Multiply by your real tariff, including everything that lands on the bill, not the discounted daytime rate.
5. Apply a downtime and reject correction. Nobody runs at 100% uptime. Take the uptime you actually expect and multiply D by it. Compare accepted against rejected shares for the last day in the pool dashboard, and treat a gap above 5% between live and rated hashrate as a hardware or link fault rather than a pool problem.
6. Build in falling revenue from difficulty growth. This is the step almost everyone skips. Difficulty retargets every 2016 blocks, roughly every two weeks. Your hashrate stays put, so your share of the network shrinks and BTC revenue with it. Do not compute D times 365. Sum two-week chunks where D is slightly smaller each time.
7. Factor in your payout scheme. FPPS pays a fixed rate per share regardless of block luck. PPLNS pays from actual blocks found, so daily income swings. Over months the expected value converges, but short-term PPLNS variance can push your first meaningful payout weeks later than an FPPS equivalent. Pool fees on FPPS run 2.5% to 4% across major pools; PPLNS fees are typically lower (0.9% to 1%) but come with that variance. Current fees are in the pool comparison.
8. Add withdrawal fees and exchange spread if you are measuring payback in fiat. The payout threshold does not take money, it holds it: at 100 TH/s a 0.005 BTC threshold takes roughly 100+ days to fill, a 0.001 BTC threshold about 21 days.
9. Redo the sum after the next retarget. A forecast without recalculation has a shelf life of two weeks. After that it is not a forecast.

Worked comparison: three models at September 2026 numbers

All inputs from the Hashrate Index snapshot of 14.09.2026 and the retarget of 19.09.2026: network hashrate 7-day SMA 943 EH/s, difficulty after retarget 132.76 T, block reward 3.125 BTC, hashprice $40.93 per PH per day (post-retarget spot, 21.09.2026). Manufacturer specs and official store prices checked 23.09.2026. S21 Pro street price from OneMiners, 28.09.2026 (Bitmain store sold out).

Daily revenue per TH at $40.93/PH/day hashprice, after a 4% pool fee: $40.93 / 1000 * 0.96 = $0.03929 per TH per day.

Payback at each tariff (using available prices):

The M60S++ has the shortest payback at $0.05 despite lower efficiency, because its purchase price is less than half the S21 XP's. That is the whole point of running the formula rather than reading spec sheets. Price per terahash and efficiency together decide payback. Neither one alone does.

These are single-snapshot numbers. They assume difficulty stays at 132.76 T for the entire payback window, which will not happen. Real payback runs longer.

Why does an online calculator show a shorter payback than reality?

Because it holds difficulty flat. The calculator takes today's daily revenue and divides the hardware price by it, as if the network froze for the whole payback period. Difficulty retargets around 26 times a year, usually upward. After the 19.09.2026 retarget came in at +4.16%, spot hashprice moved from $39.36 (31.08.2026, Hashrate Index) to roughly $37.8 on difficulty alone (BTC price then pushed it back to $40.93). Every positive retarget stretches your payback period, and the calculator does not know about retargets that have not happened yet.

How much does difficulty growth stretch payback?

Take an illustrative scenario: BTC revenue drops 2% per retarget (not a forecast, just arithmetic for showing the mechanics). After 13 retargets, about six months, the S21 XP's daily gross falls from $10.61 to roughly $8.15. Electricity at $0.05 stays at $4.38. Net drops from $6.23 to $3.77, a 39% decline. The denominator shrank, so the payback period grew.

We publish no recommended difficulty growth figure. The current pace is visible on the live difficulty chart. The 19.09.2026 adjustment was +4.16%, and the 30-day SMA of network hashrate on 14.09.2026 was 928 EH/s. Use the trend you can see, not a number from this or any other article.

Which matters more for payback, the pool fee or the electricity rate?

The rate, by a wide margin. On the S21 XP at 270 TH/s, the difference between $0.05 and $0.12 per kWh is $6.12 a day. The difference between a 0.9% pool (Tpool, HeroMiners) and a 4% pool (AntPool, F2Pool, ViaBTC, Binance Pool) on the same machine is about $0.33 a day. Almost twenty times apart. Shopping pools for half a point while your tariff assumption is wrong accomplishes nothing.

That does not make the pool irrelevant. On a thin margin, when a machine hovers near zero net, the payout scheme and the threshold decide whether you see your money in three weeks or three months. Current fees and schemes across pools are in the pool comparison.

The secondhand trap: ROI on a used ASIC

A used machine at half price does not mean half the payback period, because the denominator might be worse. Older hardware runs at higher W/TH, so it burns more electricity per terahash. An Antminer S19 XP at 140 TH/s and 21.5 W/TH costs roughly $298 median on the secondhand market. That is $2.13 per TH.

At $0.05/kWh, the S19 XP's daily electricity is 140 * 21.5 * 24 / 1000 * 0.05 = $3.61. Gross revenue at $40.93/PH hashprice after 4% fee: 140 * 0.03929 = $5.50. Net: $1.89. Payback on $298: 158 days. That looks appealing.

At $0.08/kWh, electricity climbs to $5.78, net drops to minus $0.28, and the machine loses money every day it runs. The shutdown tariff for 21.5 W/TH hardware at current hashprice is about $0.07/kWh. Above that, the purchase price is irrelevant because the machine cannot earn it back.

The lesson: check the shutdown tariff first (it is in the shutdown threshold breakdown), then run the payback formula. Not the other way around.

Should payback be measured in BTC or in dollars?

Track both, but keep them apart. BTC shows how the hardware is doing: how many coins it produces and how that number erodes as difficulty climbs. Dollars show the hardware and the market together. If BTC revenue holds steady while the dollar figure sags, the cause is price, not your farm.

Does solo mining change the payback picture?

It changes the distribution, not the total. Expected value is about the same, but the outcome becomes a lottery: nothing for a very long time, then a whole block, or nothing at all. For a small farm's payback math that is unusable, because payback needs a predictable daily flow.

Pre-purchase checklist

1. Count the entry price in full: shipping, duties, and electrical work included.
2. Use your actual tariff from the bill, not the advertised one.
3. Check the shutdown tariff for that model's W/TH at current hashprice. If your rate is above it, stop here.
4. Re-measure consumption at the wall once the unit is installed. The datasheet figure was captured under the manufacturer's conditions.
5. Assume realistic uptime, not 100%.
6. Compute in two-week chunks with revenue stepping down each time, using the pace from the difficulty chart.
7. Compare the result against secondhand prices for the same model. Sometimes the fastest payback is the purchase you skip.
8. Run the final numbers through the profitability calculator and check the model in the miner ranking.

FAQ

Which ASIC ROI calculator is the most accurate?

They all run the same formula, so accuracy comes from your inputs. What differs between services is whether they pull current difficulty, let you pick a payout scheme, and account for the pool fee. A projection built on somebody else's electricity rate will be wrong in any calculator. The POOL BTC calculator compares FPPS and PPLNS on identical inputs.

How long does ASIC payback take in 2026?

At current BTC price and $0.05/kWh, top-efficiency hardware (15 W/TH class) pays back in roughly 12 to 18 months if difficulty growth stays moderate. Older hardware at 22+ W/TH can take 24 to 36 months or may never pay back. Always model with difficulty growth scenarios, not flat projections.

Should resale value go into the payback calculation?

Yes, if you plan to sell. The numerator then becomes X minus the price you expect to get within your horizon, not the full X. For a machine you will definitely sell in two years, the full X overstates the payback period.

What if the calculation comes out negative?

Do not buy at that tariff. A negative denominator means the machine grows its loss every day it runs, and no pool switch fixes that: the fee spread across pools is measured in cents and the shortfall in dollars. The options are cheaper energy, a more efficient model, or staying out.

How often should payback be recalculated after purchase?

After every difficulty retarget, roughly every two weeks. While you are there, compare the worker's live hashrate in the pool dashboard against the rated figure. A gap above 5% points to hardware or connectivity and feeds straight into a longer payback.

Does the payout scheme affect ROI?

Not the total expected value over a long period, but the cash flow pattern. FPPS smooths income but charges a higher fee (typically 2.5% to 4%). PPLNS passes block luck through, so some weeks you earn more and some less. For payback math, FPPS gives a more predictable denominator. For total coins over a year, the difference is usually within a few percent.

This article contains referral links to mining pools (marked as sponsored). We may receive a reward if you register through them. This does not change the figures or the order of rows in the tables: the terms are taken from the pools' official pages.