Bitcoin mining keeps the network honest by turning electricity into cryptographic security. In 2026 the hashrate that measures that security has been swinging between roughly 850 and 1,000 EH/s, and the difficulty figure just hit a fresh record of 132.76 trillion. Understanding why “hash” moves now requires a clear look at how the work actually gets done.
Block assembly
Miners pull unconfirmed transactions from the memory pool and assemble them into a candidate block. They then build a header that includes the previous block’s hash, the Merkle root of the transactions, a timestamp, the current difficulty bits, and a counter called the nonce.
That header is run through two rounds of SHA-256, the same function used since the network began in 2009. The resulting 64-character string is compared to the current target. If the hash is smaller than the target, the block is valid and is broadcast to every other node.
If the hash is too large, the miner changes the nonce and tries again. Modern ASICs can perform trillions of these attempts every second, which is the literal definition of network hashrate.
Proof of work target
The target value is set so that, on average, one valid block appears every ten minutes. When more miners join or upgrade hardware, the average interval shortens, so the protocol raises the difficulty every 2,016 blocks, roughly two weeks.
The September 19, 2026 adjustment lifted difficulty by 4.16 percent to 132.76 trillion. That single number tells operators how many hashes they must produce before any one of them is likely to solve the puzzle.
Because the adjustment is automatic, the network can absorb sudden influxes or exits of hash power without manual intervention or trusted parties.
Subsidy after the halving
Each new block still pays 3.125 BTC to the winning miner, the amount set by the April 2024 halving. At 144 blocks per day, daily issuance sits near 450 BTC, plus whatever transaction fees the block contains.
Transaction fees have become a larger slice of revenue because the subsidy is smaller. When fee pressure is light, the revenue per unit of hash, called hashprice, can drop below the cost of electricity for older machines.
Publicly traded miners reported average production costs near $75,500 per coin in the middle of 2026, a figure that explains why some fleets were idled and why total hashrate showed occasional dips even as newer rigs came online.
Hardware efficiency gains
New ASICs have narrowed the gap between revenue and cost. The Bitmain Antminer S23 Hyd 3U reaches 1,160 TH/s at 9.5 joules per terahash under hydro cooling. Auradine’s Teraflux series, shipping in volume this year, lists 9.8 to 11 J/TH across air, hydro, and immersion models.
These numbers matter because each incremental improvement in joules per terahash lets an operator stay profitable at higher electricity rates or lower hashprice. The result is visible in the network: hashrate recovered toward 950 EH/s in September even after earlier economic pressure.
Older S19-era machines that needed 29 J/TH or more are being retired or sold into secondary markets, shifting the fleet mix toward the new efficiency frontier.
Hashrate volatility
Network hashrate is simply the sum of every miner’s output. In 2025 it briefly exceeded 1,150 EH/s. In 2026 the seven-day average has moved between 850 and 1,000 EH/s as operators react to price, weather, and alternative uses for their power contracts.
Some facilities curtailed hashing to sell power or capacity to AI workloads, an option that became attractive when Bitcoin revenue per hash fell. Those decisions appear in real time on public dashboards that track total SHA-256 hashes per second.
Because difficulty follows hashrate with a two-week lag, a sudden drop in hash power lowers the target and makes the remaining miners slightly more profitable until the next adjustment.
Security implications
Higher aggregate hash power raises the cost of any attempt to rewrite recent blocks. An attacker would need to out-compute the honest network for the duration of the attack, an expense measured in electricity and hardware at global scale.
The 2026 difficulty peak of 132.76 trillion therefore functions as a live barometer of how expensive such an attack would be today. Even when individual mining companies report losses, the network’s overall security metric remains elevated.
That separation between miner profitability and network strength is built into the protocol; difficulty only cares about total hashes, not who produced them.
Fee market role
Block space demand sets the fee portion of miner revenue. When mempool backlogs grow, users bid higher to be included, and those fees can offset a lower subsidy. In periods of low on-chain activity, fees shrink and hashprice suffers.
Operators watch both the spot price of Bitcoin and the fee market when deciding whether to keep machines online. A rally that lifts Bitcoin from the mid-70,000s into higher ranges quickly improves the economics even if difficulty also rises.
The September 2026 difficulty jump coincided with one such price move, illustrating how market signals travel through hashprice to hashrate and finally to difficulty within a single adjustment cycle.
Geographic and corporate mix
Public companies such as Marathon, CleanSpark, and Bitdeer together represent a sizable share of current hash power. Their quarterly disclosures give the market an early read on fleet utilization and expansion plans.
Private pools and smaller operators still account for the majority of hash, but the listed firms’ decisions on curtailment or expansion often move the visible metrics first. Their disclosures in 2026 showed both equipment cancellations and new deployments of hydro-cooled rigs.
Regional factors, from Texas weather alerts to Canadian hydropower contracts, continue to influence where hash power is located and how steadily it runs.
Next adjustment cycle
The next difficulty change is already set in motion by whatever hashrate prevails over the coming fortnight. If new machines keep shipping and Bitcoin holds above production costs for the efficient fleet, the figure is likely to test higher records.
Should a sustained drop in price or a wave of AI-related curtailments remove hash, the subsequent adjustment will ease the target and restore margin to the remaining operators. Either outcome leaves the core mechanism unchanged: every 2,016 blocks the protocol recalibrates so that blocks still arrive roughly every ten minutes.
That automatic response is why the single word “hash” now functions as shorthand for both the work being performed and the security level the network currently enjoys.
Forward look
Bitcoin mining in 2026 shows a network whose security metric tracks hardware efficiency and real-time economics rather than any single company’s balance sheet. As long as SHA-256 hashing remains the required proof, the hashrate figure will continue to register the outcome of those forces every two weeks.
