The proof-of-work hashrate leaderboard doesn't exist
Twenty-eight cryptocurrencies still carry a hashrate figure. Between them they use twenty different algorithms. A single ranking that adds them all up is not a ranking of security, of work, or of anything else — it is a ranking of ticker symbols that happen to report a number.
The intuition behind a hashrate leaderboard is straightforward: a chain does X hashes a second, so put them in descending order and the biggest chain wins. Every aggregator publishes one. Almost none of them mean what a reader thinks they mean.
Same suffix, different unit
Hashrate reads as if it were a physical quantity — hashes per second, a rate, comparable across anything that reports it. It isn't. A SHA-256 hash and a Scrypt hash are different pieces of work wearing the same suffix, in the same way that a joule of electricity and a joule of a peanut are both energy and are not equivalent things to compare.
SHA-256 is deliberately cheap: an ASIC performs some 1014 of them a second. Scrypt is memory-hard and each hash costs far more silicon and energy. So Dogecoin at a few PH/s and Bitcoin Cash at a few EH/s are not separated by six orders of magnitude of security — they are priced in different units. Comparing across algorithms needs cost — energy burned or dollars spent per unit time — not hash counts.
Bitcoin, and then everyone else in its algorithm
SHA-256 is the largest algorithm and it has five chains. Bitcoin holds most of it.
| Algorithm | Chains | Leader |
|---|---|---|
| SHA-256 | 5 | Bitcoin — 917 EH/s |
| Scrypt | 2 | Litecoin — 2.46 PH/s |
| RandomX (CPU) | 2 | Monero — 5.93 GH/s |
| Equihash | 2 | Zcash — 25 GH/s |
| Multi-algorithm | 2 | DigiByte — 45 PH/s (summed across five) |
| Etchash | 1 | Ethereum Classic — 147 TH/s |
| kHeavyHash | 1 | Kaspa — 309 PH/s |
| X11 | 1 | Dash — 3.63 PH/s |
| KAWPOW | 1 | Ravencoin — 774 GH/s |
| Eaglesong | 1 | Nervos Network — 71 PH/s |
Ten of the twenty algorithms have one chain. The other ten have between two and five. There are twenty tables here, not one.
The trap the aggregators fall into
If you sort all twenty-eight by their raw hashrate figure, the second entry — above every Bitcoin fork, above Kaspa, above every other proof-of-work chain — is Pearl, an 82-million-dollar chain that publishes a figure of roughly 24 EH/s from a consensus it calls Proof-of-Useful-Work. The number is the network's self-report and the algorithm has no external comparison, which is a warning label, not a ranking.
An aggregator that sorts by the raw column prints that second place because the column exists, and treats the reader's confusion as somebody else's problem. A ranking that groups by algorithm doesn't have the option — the reader sees at a glance that Proof-of-Useful-Work has exactly one chain and no meaningful league to lead.
Dogecoin's real number is zero
Dogecoin's Scrypt hashrate is real work, but it is not Dogecoin's work. Every block sampled on the chain carried an AuxPoW header, meaning it was won by a miner being paid to secure Litecoin (or another Scrypt parent) and merely accepting Dogecoin's block subsidy along the way. The chain's own hashrate — the amount of work that would still be done if every Scrypt parent went offline tomorrow — is nil.
The borrowed number is real security, in the sense that a miner willing to double-spend Dogecoin has to outspend the whole Scrypt fleet to do it. It is simply not the chain's own work, so the column reads 0 and the borrowed figure is shown next to it, unranked.
What the ranking does about it
The hashrate column here is grouped by algorithm rather than presented as one number. Chains with no comparison — the twenty percent whose algorithm has exactly one member — are still listed, still show their figure, and are still deliberately not ranked against anyone else. A blank is a better answer than a fake number.
See it in practice on the full list of 527 cryptocurrencies, or the ranking itself. The methodology page sets out the rule.
Figures are from the data build of 22 August 2026 and move with every six-hourly refresh. Twenty-eight chains and twenty algorithms is where the ranking sits today; both numbers drift as chains halt, merge or migrate, but the shape of the argument does not.