Antminer S21 Efficiency in J/TH: A 2026 Breakdown

For Bitcoin miners, the single most important spec is no longer raw hashrate — it is efficiency, measured in joules per terahash. The Antminer S21 efficiency in J/TH is the number that decides whether the machine earns or burns at a given electricity price, especially after the 2024 halving tightened the math. Buyers searching this exact metric want to compare the S21 variants and understand what the figure means in practice. This guide breaks it down with cited specs. Coin Web Mining stocks the current S21 range as an independent reseller, and the numbers below come from Bitmain’s spec sheets.

What is the Antminer S21 efficiency in J/TH?

The base air-cooled Antminer S21 is rated at roughly 17.5 J/TH on Bitmain’s spec sheet, producing about 200 TH/s. The more efficient S21 XP drops into the low-teens J/TH range, and the hydro variants push efficiency further still while delivering much higher per-unit hashrate. These figures represent a major step down from the S19 era, which sat around 30 J/TH or worse.

Joules per terahash measures how much energy the machine spends to produce one terahash of computation. Lower is better: a 13 J/TH unit produces the same hashrate as a 26 J/TH unit while drawing half the power. That ratio is the entire reason the S21 generation stays profitable on electricity prices that would sink older hardware. The full series context sits in the Bitmain Antminer ASIC catalog.

The unit is straightforward once you see it as a rate. A joule is a unit of energy and a watt is a joule per second, so J/TH is just the power draw in watts divided by the hashrate in terahash. A machine pulling 3,500 W at 200 TH/s is spending 17.5 joules for every terahash, every second. Because the figure normalizes power against output, it lets you compare a 200 TH/s unit and a 100 TH/s unit on equal footing — the smaller machine is not automatically worse, and the larger one is not automatically better. What matters is which one spends fewer joules per terahash.

Efficiency by S21 variant

The S21 line spans several efficiency tiers, and the differences are large enough to change the buying decision.

Air-cooled S21 and S21 XP

The base air-cooled S21 sits near 17.5 J/TH. The S21 XP improves into the low-teens J/TH range, making it the efficiency choice for grid-rate operators who run on standard power without water cooling.

Hydro and immersion variants

Hydro units such as the S21 XP Hyd use liquid cooling to run the chips harder while holding efficient sub-15 J/TH figures, and they pack far higher hashrate per machine. Immersion variants achieve similar gains in dielectric-fluid tanks. Confirm the exact J/TH for any specific SKU against the manufacturer spec sheet, since each configuration differs.

One nuance trips up buyers comparing variants: the same chip can post different J/TH numbers depending on its power mode. Firmware often exposes an efficiency mode, a balanced mode, and a high-performance mode. Efficiency mode lowers the clock to spend fewer joules per terahash at the cost of some hashrate, while high-performance mode does the reverse. So a single physical unit does not have one J/TH figure but a small range, and the right setting depends on whether your priority is squeezing margin from expensive power or maximizing output on cheap power.

Why J/TH decides profitability in 2026

Mining revenue is fixed by the network: every miner earns a share of block rewards proportional to its hashrate, paid in Bitcoin whose price moves daily. Cost, however, is set largely by efficiency multiplied by your electricity rate. Two machines with identical hashrate but different J/TH earn the same revenue and pay different power bills — and that gap is the margin.

The 2024 halving cut the block subsidy to 3.125 BTC, roughly doubling the efficiency needed to break even at any given power price. Network difficulty keeps climbing, and hashprice — daily revenue per terahash — shifts weekly. An efficient S21 keeps more of each revenue dollar, which is why the J/TH figure is the first number a serious buyer checks. Our post-halving economics guide shows how efficiency feeds the break-even calculation.

It is worth being explicit about why revenue is fixed but cost is not. The Bitcoin network pays out a set amount of new coin per block regardless of how efficient any individual miner is, and each miner earns a share proportional to its hashrate. So two miners with identical hashrate earn identical revenue no matter their efficiency. Their costs diverge entirely on the power side. The efficient one pays a smaller electricity bill for the same income, and that smaller bill is the margin. Efficiency does not raise revenue — it lowers cost, which in a fixed-revenue system is the same thing as raising profit.

How to use J/TH when choosing a unit

The practical method is simple. Multiply the unit’s J/TH by your electricity price to estimate the energy cost of each terahash, then compare against current hashprice. If hashprice per terahash exceeds your energy cost per terahash, the machine earns before hardware amortization; if not, it loses. A lower J/TH widens the band of electricity prices at which the unit stays profitable.

This is why a buyer on expensive grid power should prioritize the lowest-J/TH variant they can afford, while a buyer on very cheap power has more latitude. Run the comparison against live data, not a fixed snapshot, because hashprice and difficulty both move. The Antminer S21 variants comparison lines up the J/TH figures across the family to support that choice.

Consider two buyers to see the logic. One pays twelve cents per kilowatt-hour in a typical residential market; for them, every reduction in J/TH directly widens a thin margin, so paying up for the most efficient XP variant is usually worth it. The other has a two-cent industrial contract; their electricity is so cheap that the efficiency difference between variants barely moves the bottom line, so they may rationally favor a higher-hashrate unit even at slightly worse J/TH. Same metric, opposite optimal choice — because J/TH only becomes money once you multiply it by your specific power rate.

Beyond J/TH: what else affects real efficiency

Rated efficiency assumes ideal conditions. Real-world J/TH can drift higher with high ambient temperatures, dust buildup, or aging thermal paste, all of which force the unit to work harder or throttle. Firmware also matters: some operators tune for efficiency over raw hashrate, trading a little output for a better J/TH on cheap-but-not-free power.

Power-supply losses and PDU overhead add a few percent to the wall figure beyond the chip-level rating, so model with the at-the-wall number. Coin Web Mining is an independent reseller on a 1–3% margin, not a Bitmain partner, so the guidance here is to read the manufacturer J/TH spec carefully and verify real conditions rather than trust an idealized label.

Ambient temperature deserves particular attention because it is the variable buyers most often ignore. A unit that posts its rated J/TH in a 20°C room can drift meaningfully worse in a 35°C summer space, since the chips run hotter and either draw more power or throttle to protect themselves. Operators in warm climates effectively pay an efficiency tax that the spec sheet, measured under controlled conditions, never shows. Planning for cooling — or siting hardware where ambient temperatures stay moderate — is part of preserving the efficiency you paid for, not an optional extra.

Finally, treat the manufacturer figure as a starting point rather than a guarantee. The published J/TH reflects a healthy unit at a specified power mode in ideal conditions. Your real number depends on firmware setting, ambient heat, hardware age, and the efficiency of the power supply feeding it. Building a profitability model on the at-the-wall figure under your actual conditions, rather than the headline spec, is what keeps the math honest.

This matters especially for buyers in warm regions such as Singapore and Southeast Asia, where high ambient temperatures push real efficiency above the lab figure year-round. The same S21 that holds its rated J/TH in a temperate data center may run measurably worse in a tropical room without active cooling. Factoring that local reality into the model — rather than assuming the spec-sheet number — is the difference between a forecast that holds up and one that quietly erodes once the hardware is running.

The takeaway is to read J/TH as the single most decision-relevant spec, but to treat the published number as a best case rather than a promise. Verify it against your firmware setting, your ambient temperature, and your power-supply efficiency, then multiply by your real electricity rate to see where the unit lands relative to live hashprice. That sequence converts a marketing figure into an honest profitability estimate, which is the only version of the number worth basing a purchase on.

References

• Bitmain Antminer S21 efficiency specifications — Bitmain
• S21 variant J/TH and hashrate comparison — ASIC Miner Value
• Hashprice index and efficiency-driven economics — Hashrate Index
• Live Bitcoin difficulty and network hashrate — mempool.space

To match an efficiency tier to your power rate, the Coin Web Mining catalog lists the current S21 range with live pricing — or request a quote for bulk orders.