How proof-of-work mining actually works
A miner takes a block header (prev hash, transactions, timestamp) and repeatedly changes a , hashing the whole thing with SHA-256 until the output is below the network target — i.e. it starts with enough leading zeros. The first miner to find a valid hash broadcasts the block and earns the plus fees. The network’s retargets periodically so the average block time stays constant regardless of how much joins or leaves.
One digit, a whole new hash
Open the Hash Hunt lab → Race to find an input whose SHA-256 starts with N zeros and feel how each extra zero multiplies the work — that’s the core of PoW difficulty.
The arms race: CPU → GPU → FPGA → ASIC
- CPU (2009) — Satoshi mined the genesis block on a regular CPU.
- GPU (2010–11) — graphics cards parallelize hashing far better than CPUs.
- FPGA (2011) — field-programmable chips beat GPUs on efficiency.
- (2013–present) — application-specific chips built for one algorithm dominate. Bitcoin is now mined almost exclusively on SHA-256 ASICs.
Why ASICs dominate Bitcoin: SHA-256 is simple and stable, so a chip built only to do SHA-256 is orders of magnitude more efficient than a general-purpose computer. ASIC resistance (e.g. Ethash, RandomX) tries to make specialized hardware less advantageous so commodity hardware can still compete — survives on such algorithms (Ravencoin, pre-Merge Ethereum); Kaspa, once GPU-mined, has since moved to dedicated kHeavyHash ASICs.