ASIC describes specialization, not consensus authority. The chip repeatedly evaluates double-SHA-256 over candidate 80-byte block headers and reports results below an assigned target. A complete miner also needs a controller, power conversion, clocking, cooling, firmware and a source of block templates or pool jobs. Hashrate measures search throughput; J/TH measures electrical energy per terahash; neither guarantees profitability or control over Bitcoin's rules.
An ASIC implements a narrow workload directly in silicon instead of spending transistors on general-purpose instruction handling. Bitcoin mining therefore moved from CPUs and GPUs through FPGAs to SHA-256 ASICs because fixed-function datapaths can perform vastly more hashes per joule. The trade-off is inflexibility: a SHA-256 mining ASIC cannot simply be repurposed into a general computer when the mining workload changes.
Bitcoin proof of work hashes the serialized 80-byte block header twice with SHA-256. The header contains version, previous-block hash, Merkle root, timestamp, nBits and a 32-bit nonce. Mining silicon is optimized around this compression workload; implementations can reuse intermediate state when most header bytes stay constant, but the network validates only the resulting header hash against the target, not the miner's internal optimization.
A 32-bit nonce provides only 2^32 direct values, far too small for modern hashrates. Mining systems therefore alter additional degrees of freedom: coinbase extranonce changes the coinbase transaction and therefore the Merkle root; nTime can be rolled within consensus limits; BIP320 version bits may extend header search space. Stratum V2 explicitly assigns unique search-space regions so devices do not waste energy duplicating the same candidates.
The ASIC does not discover transactions or consensus rules by itself. A full node or template provider constructs a candidate block, historically exposed through getblocktemplate under BIP22 and pooled-mining extensions in BIP23. Pool or proxy software then transforms that template into mining jobs and supplies changing header fields, targets and identifiers to devices. Stratum V2 separates Mining, Job Declaration and Template Distribution roles so work distribution and transaction selection need not be the same function.
Pools normally give miners a share target much easier than the Bitcoin network target. A returned hash below the share target proves a measurable amount of work for payout accounting; only the much rarer result below the network target is a valid block candidate. Pool difficulty therefore changes how often shares are submitted, not Bitcoin's consensus difficulty or the miner's physical hashrate.
Hashrate is hashes per second and describes throughput. Electrical input is measured in watts, while efficiency is commonly expressed as joules per terahash, J/TH = watts divided by TH/s. Lower J/TH means less electrical energy per unit of search. Profitability still depends on Bitcoin price, network difficulty, fees, uptime, electricity, cooling, financing and hardware depreciation; a more efficient machine can still lose money at an unfavorable power price.
Real chips from the same design are not identical because semiconductor process variation affects leakage, timing and voltage margin. Firmware can trade frequency and core voltage against hashrate, efficiency and stability; underclocking often improves J/TH while overclocking may increase hashrate at disproportionately higher power and heat. Autotuning systems characterize individual chips or hashboards rather than assuming one frequency-voltage point is optimal for every die.
A mining appliance is a power-and-thermal system around the ASICs. High-current regulators, connectors, hashboards, fans or liquid loops and the controller all constrain usable hashrate. Excess junction temperature, inadequate airflow, dust, fan failure, unstable supply voltage or aggressive clocks increase hardware errors and shutdown risk. Nameplate ASIC efficiency is therefore not identical to wall-plug efficiency because PSU, control and cooling losses also consume energy.
ASICs raise the capital and engineering threshold for competitive mining, creating economies of scale in procurement, energy, repair and infrastructure. That can concentrate manufacturing or hashpower, but concentration is not identical to consensus ownership: miners choose which valid blocks to extend while full nodes independently decide validity. Security analysis should separate ASIC supply-chain concentration, mining-pool concentration, physical site concentration and actual ownership of hashpower.
An ASIC can search candidate headers and, through its upstream software, choose which job to work on. It cannot make an invalid block valid, change the 21-million supply rule, spend coins without signatures or force a full node to adopt new consensus rules. Operators should verify firmware provenance, pool endpoint, reported versus pool-side hashrate, rejected-share rate, power at the wall, temperatures and the block-template policy they intend to support.
For the clearest picture, read this entry together with Mining, Hashrate, Difficulty adjustment, Braiins, Bitcoin, SHA-256. The reverse links also lead from Hashrate, Braiins, Heat Reuse Mining, ASIC Efficiency.