Ethereum Classic (ETC): a full reference on the coin and its mining

Ethereum Classic is the continuation of the original Ethereum chain, the one that in July 2016 refused to roll back the consequences of The DAO hack. Ethereum ran a hard fork at block 1,920,000 on 20 July 2016 and returned the lost funds through an out of band change of state. Part of the community treated that as a breach of the main rule of the network and stayed on the old chain. That is how ETC came about.

The idea at the heart of the project is an immutable history: what is written into a block stays in the block, and the rules change only through the open ECIP proposal process. Technically this is still the Ethereum virtual machine, which means Solidity smart contracts, addresses of the 0x form and the familiar developer toolset. The difference is in consensus: ETC stayed on proof of work at the moment Ethereum moved to proof of stake in September 2022.

For a hardware owner this is today the largest network of the Ethash family by hashrate. After Ethereum switched to PoS a sizeable share of the freed capacity came here, and since then ETC has been the main home for Etchash machines. But on 22 July 2026 revenue per unit of hashrate dropped by 20 percent: the sixth era began and the block reward fell from 2.048 to 1.6384 ETC. The price was sliding at the same time, so nothing compensated for it.

This page is written as a reference: network parameters, protocol history, a ranking of the four Etchash models in our catalogue, pools, the economics on European tariffs, an explanation of how the DAG file grows, and a separate section on the numbers that the big aggregators still publish in an outdated form.

• 2016The DAO hack in June. On 20 July, at block 1,920,000, Ethereum runs a hard fork and returns the funds through an out of band change of state. The chain that refused the rollback carries on as Ethereum Classic and keeps the original history in full.
• 2017The Die Hard upgrade at block 3,000,000 on 13 January adds replay protection and finally separates the two networks. On 11 December, at block 5,000,000, the Gotham upgrade introduces ECIP-1017: a fixed supply cap and a reward cut of 20 percent every 5,000,000 blocks. The reward falls from 5 to 4 ETC.
• 2018On 29 May, at block 5,900,000, the difficulty bomb inherited from Ethereum is removed from the protocol for good. The network stops depending on external delays and gains a predictable block time.
• 2019On 5 and 7 January the network goes through two 51 percent attacks in a row. In both cases the target is a crypto exchange, and losses run into millions of dollars in double spends. Exchanges raise their confirmation requirements sharply. On 12 September the Atlantis upgrade at block 8,772,000 brings the network up to the Byzantium level.
• 2020On 12 January the Agharta upgrade at block 9,573,000, on 1 June the Phoenix upgrade at block 10,500,839, after which ETC is fully compatible with Ethereum at the Istanbul level. In March, at block 10,000,001, the third era begins and the reward falls to 3.2 ETC. In August the network goes through three more 51 percent attacks in a single month, the worst run in its history.
• 2020On 28 November, at block 11,700,000, Thanos activates, also known as ECIP-1099. The Etchash epoch length grows from 30,000 to 60,000 blocks. The DAG file shrinks by roughly half at once and starts growing twice as slowly. That brings 3 and 4 gigabyte graphics cards back into the network and pushes the memory death of hardware years into the future.
• 2021On 23 July, at block 13,189,133, the Magneto upgrade brings the network to the Berlin level: new transaction types and revised pricing for storage access operations.
• 2022On 25 April, at block 15,000,001, the fourth era begins and the reward falls to 2.56 ETC. On 12 February, at block 14,525,000, the Mystique upgrade brings the network to the London level, but without the EIP-1559 mechanism: ETC still has no base fee and no burning.
• 2022On 15 September Ethereum moves to proof of stake. The graphics cards freed by that move lift the ETC hashrate by roughly 71 percent within hours, the network then breaks the 200 TH/s mark for the first time and sets an all time high. ETC becomes the leading network of the Ethash family.
• 2023An infrastructure rebuild: the public Ethercluster endpoints go dark on 1 July and Rivet becomes the officially recommended RPC provider. Old wallet connection guides stop working, and some reference sites still repeat them.
• 2024On 5 February, at block 19,250,000, the Spiral upgrade activates: the network gains Ethereum equivalence at the Shanghai level. This is an important date and it is regularly misfiled as 2023. In June, at block 20,000,001, the fifth era begins and the reward falls from 2.56 to 2.048 ETC.
• 2025On 4 July the ECIP-1111 to 1114 proposals go out for open community discussion. That is how the Olympia package takes shape: a fee market on the EIP-1559 model, but with the base fee routed to a protocol treasury rather than burned, plus a mechanism for funding development.
• 2026On 15 January the Olympia upgrade is announced officially, on 15 February multi client compatibility testing starts, covering Core-Geth, Besu and the new Fukuii client. On 30 March the community discusses the future fee model. Mainnet activation was not scheduled as this page was prepared, with a stated target of before 2027.
• 2026On 22 July, at block 25,000,001, the sixth era begins: the block reward falls from 2.048 to 1.6384 ETC. The block timestamp is 17:13 UTC. This is the fifth cut under ECIP-1017 and the sixth reward value in the history of the network. The coin price is falling at the same time, so nothing offsets the drop in miner revenue.

Ethereum Classic does not keep a list of unspent outputs the way Bitcoin does, it keeps global state: every address has a balance, and contracts have their own storage on top of that. A transaction is a command to change state, and the cost of running it is measured in gas. The heavier the operation, the more gas it burns, and the gas price is set by the sender in gwei.

Every block is capped by a gas limit. In ETC it sits at 8,000,000 units per block, noticeably more modest than Ethereum, and it sets the throughput ceiling directly. With a block time of about 14 seconds the theoretical maximum on simple transfers is on the order of twenty transactions per second. Actual network load runs several times below that ceiling.

Mining fixes the order of blocks. Machines compute Etchash by reading a large file held in memory and look for a result below the current target. Whoever finds a block collects the fixed reward of 1.6384 ETC, the fees of every included transaction and, with luck, a small extra payment for including an uncle block. There is no EIP-1559 mechanism in the network: no base fee and no burning exist, the whole fee goes to the miner.

Verifying a block does not take the same amount of work as finding it. A node recomputes only a few samples from the DAG file using a compact cache, then independently checks the signatures, the correctness of the reward and the resulting state. That is exactly why a miner physically cannot write extra coins to itself: a block with the wrong reward is discarded by every node without discussion.

1. The wallet signs a transactionThe key holder sets the recipient, the amount, the gas limit and the gas price, then signs the transaction. The private key never leaves the device.
2. The transaction spreads across nodesEvery node checks the signature, the balance and the correctness of the transaction number, then puts it into its own pending pool.
3. The pool builds a block templateThe mining pool picks transactions by gas price, fits them into the ceiling of 8,000,000 gas units and hands out work to the connected machines.
4. The machine computes EtchashThe hardware takes pseudorandom samples from the DAG file in memory, mixes them with the header and searches for a value that puts the hash below the target.
5. The block is publishedThe block that was found goes out to the network. The reward is the 1.6384 ETC subsidy, the fees of the included transactions and a possible extra payment for an uncle block.
6. Nodes apply the blockEvery node re-executes all transactions of the block locally, compares the resulting state with the one declared and only then accepts the block.
7. Difficulty adjustsDifficulty is recalculated after every block from the time it took to find it, so the network reacts to hashrate arriving and leaving within minutes, not weeks.

Etchash is Ethash with one constant changed. Both functions work the same way: the network deterministically builds a large data set, the DAG file, and the miner has to hold it in memory in full in order to take random samples from it. The algorithm deliberately runs into memory bandwidth rather than raw compute, so a chip without fast memory is useless here.

The difference lies in epoch length. In classic Ethash the DAG file is rebuilt every 30,000 blocks, in Etchash every 60,000 blocks since the Thanos upgrade of 28 November 2020. The starting size in both cases is 1,024 megabytes and the increment is 8 megabytes per epoch. The direct consequence: at the same chain height the ETC file is roughly half the size and it grows twice as slowly. That single change is what brought 3 and 4 gigabyte graphics cards back into the network in 2020.

Because of the memory requirement Etchash never became a purely ASIC algorithm. Specialised machines do exist here, but their advantage over graphics cards is measured in tens of times, not thousands as on SHA-256. An Etchash ASIC is essentially a densely packed set of memory with support circuitry, and such machines compete first of all on energy efficiency per megahash, not on an absolute gap.

The practical consequence for a buyer is simple and harsh: the service life of a machine is set by the amount of memory soldered onto it. You cannot overclock it, you cannot add to it, you cannot work around it with firmware. When the DAG file outgrows the physical memory, the machine does not slow down, it simply stops mining that network. That is exactly how the life of the Antminer E3 with its four gigabytes ended in 2020.

The strength of the algorithm is a long horizon: in ETC the file grows more slowly than in any other network of the family, which means hardware lives longer. The weakness is that the horizon is still finite and known in advance to within a month, and moving a machine to another coin on the same algorithm is usually impossible for exactly the same reason, memory.

Full description of the Etchash algorithm →

The supply rule is fixed in ECIP-1017 and has been in force since the Gotham upgrade of 11 December 2017. The wording is simple: every 5,000,000 blocks the block reward is reduced by 20 percent. This is not a halving: the cut is a fifth, not a half, and the interval is measured in blocks, not in four year periods. The word halving applied to ETC is incorrect, even though almost every aggregator uses it.

The eras so far: 5 ETC up to block 5,000,000, then 4 ETC from December 2017, 3.2 ETC from March 2020, 2.56 ETC from April 2022 and 2.048 ETC from June 2024. The sixth era began on 22 July 2026 at block 25,000,001 and the reward is 1.6384 ETC. The value was verified by a direct query to the chain: blocks mined in early August 2026 carry a fixed reward part of 1.6384 ETC plus a small extra payment for including uncle blocks. The next reduction happens at block 30,000,001 and gives 1.31072 ETC.

The supply cap is roughly 210,700,000 ETC, including the inheritance from the Ethereum genesis, about 72,000,000 coins. In circulation at the start of August 2026 there are a little under 158 million, and exact figures differ between sources by hundreds of thousands of coins. At the current reward and a block time of about 14 seconds the network issues on the order of 3.7 million ETC a year, so supply inflation runs at about 2.4 percent a year and steps down after every era.

There is no burning mechanism in the network. The Mystique upgrade brought London compatibility, but without EIP-1559: there is no base fee and fees go to the miner in full. The Olympia upgrade under discussion changes exactly this: a base fee will appear, but it will go to a protocol treasury rather than being burned as in Ethereum. The stated plan is that the block reward and the tips stay with miners unchanged. The activation date was not set as this page was prepared.

The contribution of fees to miner income today is close to zero. The average network fee is measured in fractions of a cent and the daily transaction count sits around ten thousand, so miner income is effectively the subsidy alone. Plan the economics from the block reward and the coin price only, and treat any calculation where fees add a visible amount as having nothing to do with ETC.

The full table of Etchash models from our catalogue, sorted by efficiency: the lower the joules per megahash, the lower the tariff at which the machine still avoids running at a loss.

Browse miners in the catalogue →

The ranking is built on real efficiency from the catalogue, not on how popular a model is. Break even thresholds are calculated from the 1.6384 ETC reward, a network hashrate of about 175 TH/s and a price of about 7 dollars, all recorded at the start of August 2026, and they count electricity only.

The situation here is fundamentally different from SHA-256. A third party firmware scene for Etchash machines effectively does not exist: the support lists of HashCore Toolkit contain no Antminer E9, no E9 Pro and no Jasminer models, and no open builds with per board tuning are released for them. The only available path is official updates from the manufacturer.

The reason is technical rather than commercial. The gain from reflashing on SHA-256 comes from fine tuning voltage and frequency of the compute chips, and from the AsicBoost trick, which saves part of the work of the hash function itself. Neither applies to Etchash: the bottleneck here is memory bandwidth, not chip logic. Firmware cannot raise it, and no trick of the AsicBoost class has been devised for memory bound algorithms.

So all the optimisation available to the owner of an ETC machine comes down to three things: keep memory temperature in range, do not lose shares on a poor network link, and pick a pool with a sensible payout scheme. There is no way to pull extra percent of efficiency out of firmware here, the way it is done on Antminer S19 and S21 series machines.

We warn separately about second hand market risks. Etchash machines sometimes arrive with homemade builds of dubious origin that deliver no gain but do contain a swapped payout address. Before putting a machine to work, check the firmware version, compare it against the official support page of the manufacturer, and reflash to the factory build at the slightest doubt.

There are two ways to start: an Etchash ASIC or a graphics card with enough memory. At the start of August 2026 the ETC DAG file takes about 4.26 gigabytes, so formally even a 6 gigabyte card is enough. But the economics of graphics cards on European tariffs are negative by a wide margin, so in practice the conversation is about specialised machines only.

The minimum kit for an ASIC: the machine itself, the factory power supply, a dedicated power line, wired internet and a pool account. You do not need to run your own node, the work comes from the pool. Draw across the catalogue models runs from 620 watts on the Jasminer X16-Q to 2,200 watts on the Antminer E9 Pro: the junior models get by on an ordinary line, the senior ones need their own breaker.

Noise and heat differ radically between models. The compact Jasminer servers drawing under a kilowatt are tolerable in a utility room, whereas the Antminer E9 and E9 Pro run at an industrial noise level and are unacceptable in living space. That is not a question of comfort but of conflict with neighbours and with local rules.

The main question before a purchase has an honest answer: on Spanish tariffs not one of the four catalogue models turns a profit. Break even thresholds sit between 0.021 and 0.054 dollars per kilowatt hour, while retail and even industrial tariffs in Spain are several times higher. Mining ETC in 2026 makes sense only where power is your own, curtailed or very cheap, or where the heat from the machine replaces heating and is counted in the calculation.

It is also worth understanding the long horizon. The DAG file grows by about 8 megabytes per epoch, that is less than a megabyte a day. The 5 gigabyte threshold will be passed at the end of 2028, 6 gigabytes around 2032, 8 gigabytes at the end of the thirties. For the catalogue machines that means memory will outlast any reasonable ownership period, and they will leave the game on economics long before that.

Income for an ETC miner comes from four multipliers: machine hashrate, network hashrate, block reward and coin price. Fees, unlike in other networks, can safely be left out of the calculation: at an average fee of fractions of a cent and around ten thousand transactions a day they add nothing visible. The only things you control directly are hardware efficiency and the price of a kilowatt hour.

The event of 22 July 2026 changed the picture at a stroke and irreversibly. The reward fell from 2.048 to 1.6384 ETC, so revenue per unit of hashrate dropped by 20 percent within a single block. Price did not compensate, quite the opposite: by early August the coin was about a third cheaper than in spring. Any payback calculation made before July 2026 needs to be redone.

A simple chain is the convenient way to count. The network issues on the order of ten thousand ETC a day at a hashrate of about 175 TH/s. Your share equals the ratio of machine hashrate to network hashrate. Multiply it by the daily issuance and by the coin price and you get gross revenue, from which you subtract the draw of the machine multiplied by your tariff. For the catalogue models that subtraction is larger than the revenue on any European tariff.

Budget for two risk factors. The first is hashrate coming back: difficulty in ETC is recalculated after every block, so a large farm arriving dilutes your share within hours, not over two weeks. The second is the next reward cut at block 30,000,001, which will take away another 20 percent. A machine running on a thin margin today becomes loss making after it, with no change in tariff.

Open the profitability calculator →

We deliberately avoid listing specific models and apps with support percentages: we have no verifiable data as of August 2026, and reference material goes stale fastest exactly here. The rule is simple: the wallet has to be able to add an arbitrary EVM network, and then ETC is connected by hand using Chain ID 61 and an RPC node address.

The main practical danger follows from the shared origin of the two networks. The private key of an Ethereum address opens the address with the same spelling in Ethereum Classic too, so it is easy to confuse the networks in a wallet. Always verify the selected network before withdrawing, not just the address: coins sent to the wrong place cannot be recovered, and exchanges rarely help in such cases.

Ethereum Classic is an asset with a long listing history, so it trades on practically every large venue, including those available in the EU. Liquidity is moderate: as measured at the start of August 2026 daily trading volume ran in tens of millions of dollars rather than billions, so large orders move the price more than they would in the top five of the market.

Regulated exchange traded products on the coin do not exist as far as our checks go. The only known instrument of that kind is a trust from a large manager, but it is precisely a trust, not an exchange traded fund: it trades over the counter and its price can deviate noticeably from the value of the underlying asset. We found no filings for a spot exchange traded fund on ETC, and claims that one exists should be treated as an error.

We sell hardware, not the coin. This page is a reference: choose the venue yourself, looking at its licence in your jurisdiction, the depth of the order book and the terms for withdrawal to fiat. Check the withdrawal address character by character and do not keep more on an exchange than you are prepared to lose.

When buying hardware the rules are different. Look at the warranty period, the presence of service in the EU and the amount of memory soldered into the specific model, because that is what defines the remaining service life. Used Etchash machines are often offered as universal across several coins, and that claim almost always needs checking.

Technically ETC is a fully featured smart contract platform. After the Spiral upgrade of 5 February 2024 the network matches the Shanghai level, so Solidity compilers, development environments and Ethereum client libraries work without modification. The Olympia upgrade under discussion should bring the virtual machine to a fresher level and close the gap that has built up.

Actual economic activity in the network is modest, though. The daily transaction count sits around ten thousand, the average fee is measured in fractions of a cent, and the gas limit of 8,000,000 units per block is several times lower than in Ethereum. There is no developed decentralised finance segment in ETC, and we will not quote total value locked figures here: we have no verifiable data as of August 2026.

The real role of the network is different. ETC positions itself as a conservative proof of work asset with a hard supply limit and an immutable history. It is the only large network that supports smart contracts and stays on mining at the same time, which makes it a focal point for those who reject the Ethereum move to proof of stake on principle.

The difference from Ethereum is worth holding in mind as a whole: there is no proof of stake and no staking here, no fee burning and no EIP-1559 model, no ecosystem of layer two solutions, and the supply cap is fixed. In exchange there is mining, a predictable monetary policy and a commitment never to rewrite history under any circumstances.

Ethereum Classic has the heaviest history of 51 percent attacks in the industry. The first two happened on 5 and 7 January 2019, three more in August 2020. In every case the target was a crypto exchange: the attacker deposited coins, swapped them and withdrew, then rewrote the recent part of the chain and got the original amount back. Losses ran into millions of dollars.

The cause of the weakness was structural: the network ran on the same algorithm as Ethereum but had many times less hashrate, so renting superior capacity was cheap. That changed after 15 September 2022: with the Ethereum move to proof of stake the freed capacity came to ETC, the network became the largest on its algorithm, and there was nowhere left to rent comparable power. There have been no successful attacks of that kind since.

The legacy of those incidents lives on in exchange rules: confirmation requirements for ETC remain among the strictest in the industry and are counted in hundreds of blocks. For a miner that means slow turnover of what was mined, and it is worth allowing for when planning sales.

The weak spot today is not the hashrate but its distribution. Two pools control more than 85 percent of network power, so the formal attack threshold is reached by two operators colluding. Practical risk is limited because the machines belong to miners and switching to another pool takes minutes. We will not quote the number of independent nodes: no public counter with verifiable data could be found as of August 2026, and figures from old reviews are unreliable.

The block reward is published as 2.56 ETC. That value was in force from April 2022 to June 2024, so the source is two whole eras behind. At the start of August 2026 one of the largest specialist aggregators still showed this number, both in its question section on the coin and on its pool page. The error overstates calculated income by roughly a factor of one and a half, and every calculator built on it returns a wrong result. The current reward is 1.6384 ETC.

The reward is written as 2.048 ETC. That is the fifth era value, in force from June 2024 until 22 July 2026. After block 25,000,001 it is wrong. This error is more common than the previous one, because many reference sites updated once in 2024 and have not been revisited since.

The reward cut is called a halving. Under ECIP-1017 the reward is reduced by 20 percent, not by half, and it is tied to every 5,000,000 blocks, not to a four year cycle. The term halving was carried over from Bitcoin out of habit and it misleads: the real cut is two and a half times gentler than the word implies.

Era numbering and event names get mixed up. The June 2024 event at block 20,000,001 opened the fifth era, and the event of 22 July 2026 at block 25,000,001 opened the sixth. Some large venues call the 2026 event both the fifth cut and the fifth era at once, which leaves incompatible statements sitting in one text.

The date of the 2026 cut is stated wrongly. You will find 12 March 2026, as well as forecast windows from July to October and from August to October, published in spring and never updated after the event. In fact block 25,000,001 was mined on 22 July 2026, with a timestamp of 17:13 UTC.

Dates of past cuts are given with errors. Some references put the cut to 2.048 ETC on 19 February 2024, whereas it actually happened in June 2024. Such tables are worth checking by block number rather than by the calendar: the block number is a rule of the protocol, the date is merely its consequence.

The Spiral upgrade is dated to 2023. Spiral activated at block 19,250,000 on 5 February 2024. The error arose because preparation and announcements ran from the end of 2023. That date decides which Ethereum level the virtual machine of the network corresponds to.

ETHW is listed among the coins for the Antminer E9. This is the most expensive error on the page for a buyer. The EthereumPoW DAG file takes about 7,624 megabytes against 4,360 megabytes for ETC, because the epoch there stayed short, 30,000 blocks. Bitmain machines with 6 and 7 gigabytes of memory physically cannot hold it. A major pool published a direct notice: when put to work on ETHW such machines show a hashrate collapse, fail to reach nominal, reboot or do not start at all. Even so, ETHW is still listed in product cards at a number of sellers, in the official description of supported algorithms, and in calculator directories.

People assume compatibility is decided by the algorithm name. Etchash and Ethash are the same code with a different epoch length. What decides whether a machine works with a network is not the algorithm name but the DAG file size of that specific network compared with the amount of soldered memory. A statement like "it supports Ethash, so it will handle any Ethash coin" is wrong and leads to buying hardware that cannot work.

Network hashrate is quoted as one exact number. On the same day at the start of August 2026 different sources showed 168, 171.5, 175 and 184 TH/s. The spread of about 10 percent comes from different averaging windows. For planning it is correct to take a range and to remember that difficulty in ETC is recalculated after every block.

Moving to EthereumPoW is promised as insurance. The claim that a machine can always be moved to a second coin on the same algorithm is already wrong for the Antminer E9 and E9 Pro, and only partly true for Jasminer machines with 8 gigabytes: the ETHW DAG file will outgrow 8 gigabytes around the middle of 2027. This class of hardware effectively has no backup coin.

An exchange traded fund is attributed to ETC. A regulated spot exchange traded fund on Ethereum Classic does not exist as far as our checks go. What exists is a trust from a large manager that trades over the counter and can deviate from the value of the underlying asset. These are fundamentally different instruments, and substituting one for the other when judging demand is wrong.