How cryptocurrency mining actually works (and why Ethereum stopped)

Cryptocurrency “mining” in the proof-of-work sense — Bitcoin, Litecoin, Monero, and a handful of others — is a race to find numbers that produce hash outputs below a specific target. The winner adds the next block of transactions and receives the block reward. The race burns electricity by design. Ethereum, the second-largest chain, abandoned this model in 2022 because the electricity cost had become indefensible.

What miners are actually doing

Every Bitcoin block has a numeric field called thenonce. Miners try different nonces, hash the block header (which includes the nonce), and check if the SHA-256 output starts with enough leading zeros to be below the network’s current target. If yes, the block is valid and propagates to the network. If no, increment the nonce and try again.

Concrete: as of mid-2026, the Bitcoin network requires about 2⁷⁵ hash attempts on average to find a valid block. At 6 × 10²⁰ hashes per second (the current global hash rate, roughly), one block is found every ~10 minutes — that’s the protocol-targeted rate, maintained by adjusting the difficulty target every 2,016 blocks (about every 2 weeks).

Why electricity, not arithmetic

Each hash attempt is computationally cheap (SHA-256 of about 80 bytes). The work is in the volume — 2⁷⁵ attempts. Miners use purpose-built ASICs (application-specific integrated circuits) that compute trillions of hashes per second. The bottleneck is electricity: each hash costs a few microjoules, and trillions per second adds up to kilowatts per machine, megawatts per warehouse.

2024 estimates put global Bitcoin mining energy use at roughly 130-150 TWh/year — comparable to a mid-sized country’s total consumption (Argentina, ~130 TWh). Energy mix varies: ~50% renewable per Bitcoin Mining Council self-reports (industry-funded data, debated); independent estimates put renewable share at 30-40%.

The block reward and halvings

Each Bitcoin block’s winner receives:

• The block reward (newly created BTC).
• Transaction fees from all transactions in the block.

Block reward halves every 210,000 blocks (~4 years). The 2024 halving dropped it from 6.25 BTC to 3.125 BTC per block. By 2032 the reward will be 1.5625 BTC. Eventually (around 2140) the block reward goes to zero and miners will earn only transaction fees.

The diminishing reward is by design: Bitcoin’s total supply is capped at 21 million BTC. The halving schedule ensures supply grows slower over time and asymptotically stops, in contrast to fiat currencies which can be printed without protocol-level limits.

Why Ethereum left mining behind

Ethereum used a similar proof-of-work model from 2015 to September 2022, when it transitioned to proof-of-stake (the “Merge”). The motivation: energy cost. Pre-merge Ethereum used about 70-100 TWh/year — similar order of magnitude to Bitcoin. Post-merge it uses about 0.01 TWh — seven thousand times less.

In proof-of-stake, validators lock up ETH as a stake. Block proposers are selected pseudo-randomly weighted by stake. Misbehaving validators forfeit their stake. No electricity-burning race; the security comes from the economic cost of cheating (loss of the staked capital) rather than the cost of competing.

Trade-offs of proof-of-stake (debated within the crypto community):

• Concentration.Validators with more stake can be selected more often. Proof-of-work’s alternative was hashrate concentration, which had the same issue.
• Bootstrapping problem.Initial stake distribution affects long-term security. Bitcoin’s PoW had a natural fair-start property (anyone with a computer could mine in 2009); PoS chains usually have founder allocations or pre-mines that affect this.
• Recovery from attack.If an attacker dominates a PoS chain, they keep their stake. PoW attackers spend electricity that’s permanently gone.

Who still mines

Post-2022, the major remaining proof-of-work chains:

• Bitcoin (BTC). By far the largest. ASIC-mined.
• Litecoin (LTC). Scrypt-based; ASICs exist but more accessible than Bitcoin’s.
• Monero (XMR). RandomX algorithm specifically designed to resist ASICs; mineable on CPUs.
• Ethereum Classic (ETC). The pre-merge Ethereum fork that retained PoW.
• Kaspa, Dogecoin, Bitcoin Cash, etc. Smaller chains.

Can individuals still profitably mine?

For Bitcoin: effectively no at home. The hash-rate race is dominated by industrial-scale operations with $0.03/kWh electricity, $5,000 ASICs, and economies of scale at every layer. A hobbyist hash from a residential GPU contributes roughly nothing to the network and consumes more electricity than it earns.

For Monero and similar ASIC-resistant coins, hobby mining is technically viable but the rewards are small. Realistically, hobby mining is more about supporting the network than turning a profit in 2026.

The honest verdict

Proof-of-work mining is an electricity-to-block-reward machine. The model worked when networks were small and electricity costs were a fraction of block rewards. As block rewards halve and electricity prices fluctuate, the margin compresses. New chains overwhelmingly choose proof-of-stake; Bitcoin remains as the canonical proof-of-work holdout, defended by adherents on grounds of fair launch and energy-as-security.

For unit math (Wei, Gwei, ETH), see our Wei/Gwei/ETH converter and the wei, gwei, ETH explained guide. For the broader Bitcoin vs Ethereum comparison, see Bitcoin vs Ethereum.

Walkthrough: the per-block economics in 2026

Snapshot of a single Bitcoin block in mid-2026:

• Block reward (post-2024 halving): 3.125 BTC.
• Transaction fees per block: typically 0.05-0.30 BTC, spiking to 2+ BTC during congestion.
• Network hashrate: ~6 × 10²⁰ hashes/sec.
• Energy per hash on a modern ASIC (Antminer S21): ~17 J/TH (joules per terahash).
• Energy per block: 6 × 10⁸ TH per 600 seconds × 17 J = ~1.7 MWh per block at the network-aggregate level.

At $0.06/kWh industrial electricity (the rough global average for large miners), each block costs ~$100 of pure electricity at the marginal-cost level. Block reward of 3.125 BTC at $60,000/BTC = $187,500 of gross revenue. Margin is enormous in aggregate, but distributed across the entire global hashrate proportional to share. A miner with 0.01% of network hashrate earns $18.75 per block in expectation and burns $0.01 of electricity per block — a tiny absolute slice of a tiny percentage.

Common mistakes

• Confusing “mining” with “running a node”. A full node validates transactions and broadcasts them; mining additionally solves the hash puzzle to produce blocks. Running a node is cheap (any laptop), socially valuable, and earns nothing. Mining is expensive and earns the block reward only if you win the race.
• Treating hashrate as compute power. SHA-256 hashing on an ASIC is not general-purpose computation; it’s a fixed-function pipeline. A 10 TH/s ASIC cannot accelerate machine learning, rendering, or anything besides Bitcoin-family mining. “Total mining hashrate equals X exaflops” comparisons are misleading.
• Assuming PoS validators don’t earn. Ethereum validators earn ~3-5% APR on staked ETH plus MEV (miner-extractable value) tips. The transition didn’t eliminate validator income; it eliminated the electricity arms race that funded it.
• Conflating Bitcoin’s 21M cap with deflation guarantees.Lost keys (Satoshi’s ~1M BTC, lost wallets) are out of circulation permanently. Effective supply is <21M and shrinking. Whether that translates to price appreciation depends on demand, which the protocol doesn’t control.
• Treating hash rate spikes as security improvements.A doubled hashrate makes 51% attacks more expensive but doesn’t change the attacker’s probability of success at any given minority share. Security is a function of hashratedistribution, not absolute level.

For the related crypto unit math, see our wei, gwei, ETH explained guide. For the hashing primitive underlying it all, see cryptographic hashing explained.

Sources: Cambridge Bitcoin Electricity Consumption Index (2024); Ethereum Foundation Merge documentation (2022); Bitcoin Whitepaper (Nakamoto, 2008); IEA Electricity 2024 report.