Bitcoin miner efficiency: J/TH explained

J/TH and W/TH: read the units first

J/TH means joules per terahash: energy used for one trillion hashes. Divide power in W by hashrate in TH/s. A hypothetical 200 TH/s miner drawing 3,000 W works out to 15 J/TH. One watt is one joule per second, so W divided by TH/s gives J/TH.

A label written as W/TH often abbreviates watts per TH/s and has the same numeric value in miner specifications. Check that definition. Watts divided by a total number of hashes would be a different quantity. Convert units before comparing; 200,000 GH/s is 200 TH/s, and 3 kW is 3,000 W.

• Efficiency = 3,000 W ÷ 200 TH/s = 15 J/TH.
• Power = 200 TH/s × 15 J/TH = 3,000 W.
• Lower J/TH means less energy per hash; total watts describe the whole machine.

Calculate J/TH from a meter reading

Use a meter reading for one ASIC, including its power supply and built-in fans, and the local hashrate reported by that same machine. Choose a steady period after warm-up. Subtract the starting meter reading from the ending reading. Divide the kWh used by elapsed hours, then multiply by 1,000 for average watts. Divide by the average local TH/s over the same period.

For a hosted miner, ask the provider for that machine’s energy use and its local average hashrate with matching start and end times.

• Hypothetical readings, 10:00 to 11:00 UTC: 418.20 kWh at the start and 421.62 kWh at the end.
• Energy used = 3.42 kWh. Over one hour, average wall draw = 3,420 W.
• If the miner averaged 250 TH/s during that hour, 3,420 W ÷ 250 TH/s = 13.68 J/TH.

• See why the pool hashrate may differ

Same S21 XP, two different bills

Both offers list the air-cooled Antminer S21 XP at 270 TH/s and 13.5 J/TH. At 3.645 kW for 720 hours, either would use 2,624.4 kWh. The 8 November 2026 listing rates give the hosting costs below, before daily rounding. Adding only hardware gives $2,981.08 at Cyberian Mine and $3,354.84 at Hashlabs.

Verify the specifications and live offers

Bitmain specifies a typical wall draw of 3,645 W at 25°C inlet air for this variant. “Brazil Hydro” is the Hashlabs facility label in the catalog. The cooling field on both miner listings says air cooled. The catalog showed one Cyberian Mine unit at $2,800 and three Hashlabs units at $3,200 on 8 November 2026. Confirm current stock and rate protection in the listing.

The estimates use rated power and continuous operation. Actual charges follow confirmed delivery and the accepted terms. The hardware-plus-hosting figures exclude stopped-miner storage, taxes and other separate charges.

• Inspect the Cyberian Mine S21 XP listing
• Inspect the Hashlabs S21 XP listing

Hold hashrate and hours constant

For two hypothetical Bitcoin ASICs running at 200 TH/s, 15 J/TH means 3,000 W and 25 J/TH means 5,000 W. Over 24 hours, the first uses 72 kWh and the second uses 120 kWh. At the same $0.07/kWh rate, their electricity costs are $5.04 and $8.40. This isolates the effect of efficiency for equal nominal work. Hardware price, storage and other charges are separate.

When a less efficient miner costs less to run

Now compare a different pair, both at 200 TH/s for 24 hours. Miner A uses 15 J/TH at $0.07/kWh: 72 kWh × $0.07 = $5.04/day. Miner B uses 20 J/TH at $0.05/kWh: 96 kWh × $0.05 = $4.80/day. B uses more energy but spends $0.24 less per day in this simplified scenario.

The crossover rate for B is A’s rate × A’s J/TH ÷ B’s J/TH: $0.07 × 15 ÷ 20 = $0.0525/kWh. Below that rate, B has the lower energy-only cost at this equal hashrate and runtime. At that rate they are equal. This is a tariff comparison, not a recommended miner or an estimate of total ownership cost.

• Keep hashrate and hours equal before using this shortcut.
• These are hypothetical rates and machines, not current provider offers.
• Lower energy cost alone does not settle hardware price, delivered work, separate fees or provider risk.

• Check which hosting fees the comparison leaves out

Check the exact variant and measurement boundary

A model family can contain several hashrate and power variants. Check the exact variant, cooling type and operating mode. Also check whether the stated power is measured at the wall or only at the chips. Chip-only consumption and whole-machine power are different boundaries. Temperature, fans and firmware settings can change the figures. A specification is a reference under stated conditions, not a measurement of your hosted miner. Compare your result with the nameplate only after checking the operating mode, inlet temperature and whether the power figures include the same equipment.

• Use matching power and hashrate figures from the same mode and conditions.
• Do not pair one variant's hashrate with another variant's watts.
• A provider delivery percentage measures a separate question; it does not replace an efficiency measurement.
• Do not divide nameplate watts by a short pool average to diagnose efficiency; compare measurements over a matching period.

Check the offer before you sign in

Record the variant, power, hashrate, rate and billing basis. Use the calculator to test those inputs, then check that the live listing still shows the same specification, rate, availability and purchase price. Review the provider’s delivery record and rate protection before buying.

Open Mine’s provider-backed lifetime warranty and automatic hashrate failover apply while the miner remains hosted with that provider and the account is in good standing. Failover depends on available hashrate within that provider’s warranted fleet. Read the coverage and limits before creating an account.

• Test a model and hosting rate in the calculator
• Read the lifetime warranty and failover conditions