Joules per terahash (J/TH) measures how much electricity an ASIC miner uses for a given amount of hashing work. Calculate it by dividing power in watts by hashrate in terahashes per second: J/TH = W ÷ TH/s. A lower figure means less device electricity use at the same hashrate, but it does not by itself identify the best purchase or the most profitable miner.

This article compares selected SHA-256 miners by their manufacturer-stated J/TH figures, explains how to convert that number into daily electricity use, and clarifies what the metric does and does not tell a buyer or operator.

What J/TH Measures

J/TH expresses how much electrical energy a miner consumes to perform one terahash of SHA-256 computation. It is calculated as stated power in watts divided by stated hashrate in terahashes per second; because one watt equals one joule per second, the result is joules consumed per terahash of work. A lower J/TH figure means the device uses less electricity to produce the same amount of hashing work, all else being equal.

Manufacturers publish J/TH as a device specification, not as a measurement of a mining facility's total electricity use. Product sheets typically report "power on wall" or "wall-plug power" for the miner itself, measured under a stated inlet-air or inlet-coolant temperature. The measurement boundary matters: electricity supplied to the miner includes its onboard components, while external cooling equipment and site electrical losses fall outside that boundary. Treating a nameplate J/TH value as a complete facility energy figure overstates precision that the specification does not provide.

Comparing Selected ASIC Models by Efficiency

The table below summarizes manufacturer-published specifications for selected SHA-256 miners from Bitmain, MicroBT, and Canaan. Sources were reviewed on September 29, 2026. This is a comparison of selected specifications, not an exhaustive ranking or a statement of current stock availability. Figures are subject to manufacturer tolerances and the source limitations noted below.

Bitmain's S23 Hyd. Product Manual lists 580 TH/s, 5,510 W, and 9.5 J/TH. Its separate S21 XP Hyd. Product Manual lists 473 TH/s, 5,676 W, and 12.0 J/TH. Both specify typical values at 35°C inlet coolant, with ±3% hashrate and ±5% power and efficiency tolerances. The air-cooled S21 XP specifications give 270 TH/s and 3,645 W at 25°C inlet air, corresponding to 13.5 J/TH.

MicroBT's official WhatsMiner shop lists the M70S at 248 TH/s and 13.5 J/TH. The cited listing does not separately state wall power or test temperature; confirm these in model-specific documentation before treating the row as a fully specified procurement comparison.

Canaan's Avalon A15XP-209T listing contains an unresolved inconsistency: the page summary shows 3,667 W, while its detailed parameter table shows 3,720 W at the wall. The row above follows the detailed table, which also states 209 TH/s and 17.8 J/TH. Confirm the applicable specification with Canaan before purchase. Its −5°C to 35°C operating range is not an efficiency test temperature and should not be compared with Bitmain's stated test conditions.

Translating J/TH into Daily Electricity Use

J/TH becomes directly useful once converted into an estimate of daily device electricity consumption:

Device kWh/day = J/TH × TH/s × 0.024

Multiplying J/TH by TH/s gives watts; multiplying watts by 24 hours and dividing by 1,000 gives kWh/day. This formula assumes continuous operation at the stated hashrate and efficiency. It estimates the miner's own electricity use, excluding external cooling and other facility loads, and makes no adjustment for downtime.

Applying this formula to a common reference point of 100 TH/s illustrates the practical spread across the efficiency range above:

At equal hashrate, a 9.5 J/TH miner consumes roughly 46.6% less device electricity than a 17.8 J/TH miner (22.80 kWh/day versus 42.72 kWh/day). These are normalized comparisons, not claims that every listed model operates at 100 TH/s. The percentage describes device electricity use only. A full profitability model must separately account for facility electricity, operating time, applicable electricity rates, fees, and mining revenue. Check what a hosting rate or reported revenue figure already includes before adding costs or deducting fees.

Hashrate alone can also be misleading. Bitmain's stated specifications show the S23 Hyd. at 580 TH/s and 5,510 W (132.24 kWh/day at continuous stated power) against the S21 XP Hyd. at 473 TH/s and 5,676 W (136.22 kWh/day at continuous stated power). The newer model's nameplate hashrate is higher while its stated device power is marginally lower—a useful reminder that hashrate, watts, and J/TH should be reviewed together rather than selecting hardware on hashrate alone.

Why Cooling Design and Test Conditions Affect Comparisons

Published J/TH figures can be compared as manufacturer-stated specifications, provided their test conditions and measurement boundaries remain visible. Bitmain specifies its hydro-cooled S23 Hyd. and S21 XP Hyd. at a 35°C inlet-coolant temperature, while the air-cooled S21 XP is specified at a 25°C inlet-air temperature. These different conditions do not invalidate the listed specifications, but the figures alone cannot establish how much of the efficiency difference comes from cooling mode rather than chip design, operating settings, or temperature.

A hydro-cooled miner also requires compatible plumbing, pumps, and heat-rejection equipment. Electricity consumed by that external equipment belongs in the facility energy calculation; it is not directly included in the miner's published device-level J/TH.

Manufacturer tolerances add a further layer. The cited Bitmain specifications allow ±3% variation in hashrate and ±5% in power and efficiency. Canaan's detailed A15XP-209T table lists ±5% for hashrate, ±10% for wall power, and ±3% for efficiency, alongside the page inconsistency noted above. A procurement comparison should therefore treat J/TH as a manufacturer-stated value under defined conditions, not as a guaranteed field result.

What J/TH Does Not Capture

Device-level J/TH is a necessary input to a mining economics model, but it is not sufficient on its own. It does not include external cooling loads or site electrical losses; it does not reflect hardware purchase price, financing cost, or expected service life; and it says nothing about uptime, repairability, or the availability of spare parts and firmware support.

A pool dashboard alone cannot establish J/TH because it does not measure the miner's electrical input. However, pool-estimated hashrate can be paired with independently measured power for the same equipment and time window to estimate electricity use per unit of pool-observed hashing work. Share randomness, rejected or stale shares, and the averaging period affect that estimate, so it should not be treated as interchangeable with the manufacturer's device specification. Braiins provides an example of using wall power measurements and pool-side hashrate together in operational analysis.

Keep the measurement boundary and hashrate source explicit. Device efficiency, facility electricity use, pool-observed performance, and realized mining revenue answer different questions.

Monitoring Efficiency After Deployment

Once a miner is deployed, measure its average electrical input using a suitable meter or metered PDU at the miner's AC supply, and pair it with average hashrate over the same period. For a device-level check against nameplate specifications, use local hashrate and record the operating mode and inlet temperature. Calculate measured J/TH as average watts divided by average TH/s.

Use device telemetry to support diagnosis, but verify what each power field represents. A firmware power target or estimate is not automatically a wall-power measurement; for example, Braiins' monitoring documentation distinguishes measured wall power, estimated power, and power limits. Compare results with manufacturer tolerances only after accounting for differences in operating settings and conditions.

Persistent low local hashrate or unexpectedly high measured power warrants checking cooling, firmware settings, and hardware. Pool dashboards provide complementary information about connection status and share-based hashrate. If local performance is normal but pool-observed hashrate remains low after sufficient averaging, inspect connectivity and rejected shares. ViaBTC's mining pool address configuration guide covers connection setup and related troubleshooting. Reading these sources together, while retaining their different measurement boundaries, makes efficiency checks more useful.

FAQ

What does a lower J/TH figure mean for a miner?

A lower J/TH means the device consumes fewer joules of electricity to produce one terahash of hashing work, so at the same hashrate it uses less electricity than a miner with a higher J/TH figure.

Can I calculate J/TH from my pool dashboard's hashrate figure?

Not from the dashboard alone: you also need measured power. You can pair average pool-estimated hashrate with average measured wall power for the same equipment and time window, but share randomness and rejected or stale shares affect the result. Label it as an estimate based on pool-observed hashrate rather than treating it as directly equivalent to the manufacturer's device specification.

Why do air-cooled and hydro-cooled miners show different J/TH values under different conditions?

Chip design, operating settings, and temperature can all affect device efficiency. Manufacturers may specify inlet-air temperature for air-cooled units and inlet-coolant temperature for hydro-cooled units. Compare the published figures alongside those conditions; do not attribute the difference to cooling mode alone. External pumps and heat-rejection equipment add facility electricity use beyond the miner's device-level figure.

Does a higher hashrate always mean higher electricity use?

Not necessarily. Manufacturer specifications show cases where a newer model with higher stated hashrate has similar or slightly lower stated power than an older model, which is why hashrate, power, and J/TH should be reviewed together rather than hashrate alone.

Is a manufacturer's stated J/TH a facility-wide electricity figure?

No. It is a device specification measured under defined test conditions. Loads outside the miner's electrical input boundary, such as external cooling equipment and site electrical losses, must be accounted for separately.

References

• Bitmain, ANTMINER S23 Hyd. Product Manual
• Bitmain, ANTMINER S21 XP Hyd. Product Manual
• Bitmain, ANTMINER S21 XP Specifications
• MicroBT, WhatsMiner Official Shop
• Canaan, Avalon Miner A15XP-209T
• Braiins, Economics of Immersion Cooling for Bitcoin Miners
• Braiins Academy, Metrics and Labels
• ViaBTC, How to Configure a Mining Pool Address: Port and Worker Settings