In the scaling section of the Ethereum whitepaper, its 19-year-old author worried that if a blockchain grew too heavy, “only a very small number of large businesses would run full nodes”. Thirteen years later the network he described is still running, and it answered that worry in a way the paper never wrote down: not by making one chain bigger, but by selling cheap data space to other chains built on top of it. Much of Ethereum history is the story of that gap between the 2013 plan and the network of 2026.
The numbers give the scale. Ethereum’s first block was mined on 30 July 2015. On 9 November 2026 at 09:49:11 UTC the chain counter read block 26,376,540, according to a public node. The supply of ether stood at 122,202,746 ETH at 09:42 UTC the same morning, per ultrasound.money, and DefiLlama counted $134.5 billion of dollar-pegged stablecoins on Ethereum that day, more than on any other chain (Tron came second with $93.6 billion). Everything between those two dates, from a document shared in November 2013 to a network secured by staked ether, is told below, with the dates and block numbers of every major upgrade.
This is the first lesson of the Ethereum and smart contracts path. It assumes you know what Ethereum does today, which is the subject of our beginner’s guide to what Ethereum is: two kinds of account, smart contracts run by every node, gas fees paid in ether and validators instead of miners. That guide has a short list of milestones. Here we open the whitepaper itself, meet the people and the money behind the launch, and compare what was promised with what was built.
Before Ethereum: a Bitcoin writer who wanted more than payments
Vitalik Buterin discovered Bitcoin in 2011 and started writing articles about it, according to the founders page on ethereum.org. That writing led him to co-found Bitcoin Magazine with Mihai Alisie in 2012. Writing for a living about other people’s projects gave him a wide view of what the young Bitcoin community was trying to build on top of it, and of where it got stuck.
Bitcoin’s own scripting language is deliberately small. The whitepaper lists what it could not do, in four short points. It is not Turing-complete, so it cannot run loops. It is value-blind: a script cannot pay out an amount that depends on a price. It has no state beyond spent or unspent, so multi-step contracts are hard. And it is blind to blockchain data such as the timestamp or the previous block hash, a possible source of randomness. If those terms are new, our side-by-side comparison of Bitcoin, Ethereum and blockchain shows how the two networks differ today.
Projects of the time worked around those limits in two ways, which the paper describes. Some launched a separate blockchain for one job, like Namecoin, a name registration system created in 2010. Others wrote a protocol on top of Bitcoin, such as colored coins or so-called metacoins, which had to accept that Bitcoin miners would include transactions the new protocol considered invalid. Buterin’s proposal was a third way: one general-purpose chain where anyone could write the rules of their own application as a program.
The Ethereum whitepaper of 2013, read in plain language
The document is titled “A Next-Generation Smart Contract and Decentralized Application Platform” and is kept online at ethereum.org/whitepaper. Its central sentence comes early: “What Ethereum intends to provide is a blockchain with a built-in fully fledged Turing-complete programming language”. A contract, in the paper’s sense, is a program that lives at an address on that blockchain, holds money and runs exactly as written each time someone sends it a message. The rest of the paper explains how to make that safe and pay for it.
When the whitepaper was written: late 2013, revised into 2014
The date depends on which ethereum.org page you read. The official timeline of upgrades lists the whitepaper as released on 27 November 2013, and the founders page says Buterin shared it in November 2013. The banner on the whitepaper page itself says it was published in 2014. Both fit the text. It circulated from late 2013, but the version online quotes the size of the Bitcoin chain “as of April 2014” and credits a fast-block idea published in December 2013, so it kept being edited for months. When a source gives you a single date, it is picking one of those versions.
Accounts, ether and gas: the core design
Bitcoin tracks coins. The whitepaper tracks accounts. Each one has a 20-byte address and four fields: a nonce, which stops a transaction from being processed twice, an ether balance, contract code if there is any, and storage, empty by default. There are two kinds. Externally owned accounts are controlled by private keys, the wallets people hold. Contract accounts are controlled by their code, which wakes up every time the account receives a message. That split still exists in 2026.
Ether is called the “main internal crypto-fuel” of the system, the unit used to pay fees. The paper even fixes the small units: 1 ether equals 1018 wei, with szabo and finney in between. Wei is still the unit that contract code counts in, and wallets show gas prices in gwei, a billion wei.
The hardest problem of a programmable chain is the infinite loop. If anyone can run any code on every node, someone will run code that never stops. The answer is gas. Each transaction sets two fields, STARTGAS, the maximum number of computation steps it may use, and GASPRICE, the fee it pays per step. Most steps cost 1 gas, heavier ones more, and every byte of transaction data costs 5 gas. The stated aim is to “require an attacker to pay proportionately for every resource that they consume”. If a transaction runs out of gas, its changes are reverted but the fee is still paid to the miner, so spam always costs the spammer. The same logic sits behind today’s gas fees, though the fee now goes somewhere else, as the London section below explains.
Buterin also took a position on fast blocks. Short block times create more stale blocks, which helps large miners. Buterin adopted a simplified version of GHOST, a protocol introduced by Yonatan Sompolinsky and Aviv Zohar in December 2013, which counts these stale blocks, called uncles, when choosing the heaviest chain, and rewards the miners who found them. Uncles were part of Ethereum mining until 2022 and disappeared with proof of stake.
The applications the paper imagined, and what became of them
The whitepaper sorts possible uses into three groups: financial applications, semi-financial ones and applications that have nothing to do with money, such as online voting and governance. Its most quoted line is about tokens. A currency or token system, it says, is fundamentally “a database with one operation”: take X units from A, give X units to B, if A has them and has approved it. A few lines of code were enough, and that is exactly what happened later with tokens.
Tokens were only the start. The whitepaper sketches a hedging contract that pays out in dollars by reading a price from a “data feed” contract, crop insurance driven by weather data, a name registry written in a handful of lines, a decentralised file storage contract and decentralised autonomous organisations, or DAOs, groups whose rules are held in code. Here is how those ideas map onto the network of 2026.
| Idea in the whitepaper | What it became |
|---|---|
| Sub-currencies and token systems | ERC-20 tokens (standard created 19 November 2015) and dollar stablecoins: $134.5 billion of stablecoins on Ethereum on 9 November 2026 (DefiLlama) |
| Hedging contracts and stable-value currencies | Lending markets, derivatives and stablecoins of DeFi |
| Data feed contracts | Price oracles that report market data to contracts |
| Name registration | Name services that map readable names to addresses |
| Decentralised autonomous organisations | DAOs that vote on treasuries and protocol settings; the first big one, The DAO, ended in a hack in 2016 |
| Decentralised file storage | Mostly built on separate networks, not on Ethereum itself |
Nowhere does the text mention NFTs, dollar stablecoins as they exist now, ERC standards or rollups. Those came from developers after launch. Its last section explains why: “Ethereum is open-ended by design”, a base layer on which people would build things the authors had not thought of.
Money: a currency sale and a supply without a cap
The issuance plan is short. Ether would be sold for bitcoin at 1,000 to 2,000 ether per BTC, with early buyers getting the better price, to pay developers. Two further shares of 0.099 times the amount sold would go to early contributors and to a long-term reserve. After launch, miners would receive 0.26 times the amount sold every year, forever. The paper records how much was sold, 60,102,216 ETH, which shows that this section was updated after the sale of 2014.
The choice not to cap the supply was deliberate. The paper prefers “a permanently growing linear supply, as opposed to a capped supply as in Bitcoin”, arguing that it limits the concentration of wealth among early holders. Applying its own launch formula (amount sold times 1.198) gives about 72 million ETH at genesis, our computation from the paper’s numbers. On 9 November 2026 supply stood at 122.2 million ETH, roughly 70 percent more, under rules the paper did not describe: staking rewards on one side, burned fees on the other.
One sentence in that section turned out to be a forecast: “it is likely that Ethereum will switch to a proof-of-stake model for security”. The mining plan right after it is more honest still. It proposed using contract execution as the mining puzzle and called the model “untested”. That design was not the one used at launch.
From idea to organisation: Miami, the founders and the 2014 crowdsale
Buterin presented Ethereum publicly in January 2014 at the North American Bitcoin Conference in Miami, according to ethereum.org. By then a team had formed. The founders page lists eight co-founders: Vitalik Buterin, Gavin Wood, Joseph Lubin, Jeffrey Wilcke, Mihai Alisie, Anthony Di Iorio, Amir Chetrit and Charles Hoskinson.
Three of them shaped the technology that still runs today. Gavin Wood wrote the Yellow Paper, the formal specification of the Ethereum Virtual Machine, and developed Solidity, the main language for smart contracts. Jeffrey Wilcke created Geth, the original execution client and, according to ethereum.org, still the most widely used. Joseph Lubin later founded Consensys, the company behind MetaMask and Infura. Alisie helped set up the foundation in Switzerland and the legal framework of the sale.
ethereum.org dates the Yellow Paper to April 2014. The two documents do different jobs. The whitepaper is the vision, written for people. The Yellow Paper is the specification, written in mathematical notation so that independent teams could build compatible clients. A beginner reads the first. Client developers read the second.
The crowdsale opened on 22 July 2014. The announcement on the Ethereum blog set the price at 2,000 ETH per BTC for 14 days, then declining in a straight line to 1,337 ETH per BTC, over 42 days ending at 23:59 Zug time on 2 September 2014. Two weeks in, the team reported more than 25,000 BTC received for more than 50 million ETH, in 6,670 transactions; the smallest allowed purchase was 0.01 BTC and the largest single one was 500 BTC, for 1 million ether.
| Ether sale of 2014 | Figure | Source |
|---|---|---|
| Opening | 22 July 2014 | Ethereum blog |
| Length and close | 42 days, ending 23:59 Zug time on 2 September 2014 | Ethereum blog |
| Price | 2,000 ETH per BTC for 14 days, then down to 1,337 | Ethereum blog |
| After two weeks (8 August 2014) | Over 25,000 BTC, over 50 million ETH, 6,670 transactions | Ethereum blog |
| Total raised | About 31,000 BTC, about $18 million at the time | ethereum.org |
| Ether sold | 60,102,216 ETH | Whitepaper, issuance section |
In total the sale raised about 31,000 BTC, worth about $18 million at the time, according to ethereum.org. It was one of the first large token sales, and its model, sell a new token for bitcoin before the network exists, would be copied thousands of times three years later, with much less care.
The network goes live: Frontier, Homestead and The DAO (2015-2016)
Ethereum’s main network started on 30 July 2015, mid-afternoon UTC, with Frontier. ethereum.org calls it a live but bare-bones version for technical users. Blocks had a gas limit of 5,000 at first, too low for ordinary transfers, so miners could start and early users could install clients without a rush. The “thawing” came at block 200,000 on 7 September 2015, when the limit was lifted.
Homestead followed at block 1,150,000 on 14 March 2016, with protocol changes that made later upgrades easier. A pattern was set that still holds: Ethereum changes through scheduled hard forks, with new client software that switches on new rules at a fixed block or, since 2022, a fixed epoch.
Then came The DAO. Between 30 April and 28 May 2016, it sold DAO tokens for about 12 million ETH, worth about $150 million at the time, according to the SEC’s report of investigation. On 17 June 2016 an attacker began draining it and moved about 3.6 million ETH, a third of the total. A hard fork at block 1,920,000 on 20 July 2016 moved the funds to a recovery contract. A minority refused to change the history of the chain and kept the original one, now Ethereum Classic (ETC). The mechanics of that split, including the replay attacks that followed, are in our lesson on soft forks and hard forks.
The SEC came back to The DAO a year later. Its report of 25 July 2017 concluded that DAO tokens were securities, so their sale should have been registered, but it brought no enforcement action. For a whole industry selling tokens at the time, that report was the first clear signal from a US regulator.
Tokens and the ICO boom (2015-2018)
The whitepaper’s “database with one operation” became a standard. EIP-20, created on 19 November 2015 by Fabian Vogelsteller and Vitalik Buterin, defined a shared set of functions for tokens: transfer, balance, approve and a few more. Once every token used the same interface, any wallet or exchange could support a new token without new code. The name ERC-20 stuck.
In 2017 that ease turned into a boom of initial coin offerings. Teams sold new tokens directly to the public, usually for ether or bitcoin, often on the strength of a white paper alone. Crunchbase estimated that ICOs raised about $4.9 billion that year; other tallies differ, and our crypto history timeline places that boom among the other cycles of the market. Ethereum also kept upgrading: Byzantium went live at block 4,370,000 on 16 October 2017, one of several hard forks that tuned the protocol in its first years.
DeFi summer and the turn to rollups (2020)
Ethereum’s next growth spurt came from finance. DefiLlama’s historical series for the Ethereum chain puts its total value locked at about $0.93 billion on 1 June 2020, $8.15 billion on 1 September and $14.55 billion on 31 December 2020, before a peak of $107.4 billion on 9 November 2021. Those dollar figures move with the price of ether as much as with deposits. The story of lending, exchanges and yield farming in those months is in our history of DeFi.
Success made the main chain expensive. On 2 October 2020 Buterin published “A rollup-centric ethereum roadmap”. His conclusion was that the ecosystem was likely to be “all-in on rollups” for scaling in the near and mid-term. A rollup runs transactions on its own chain and posts compressed data and proofs back to Ethereum, which then provides settlement and security. The difference between a base chain and these networks is explained in our guide to layer 1 and layer 2. Nothing like this appears in the whitepaper, whose scaling section discussed state trees and light clients.
Two months later, proof of stake started in parallel. The staking deposit contract was deployed on 14 October 2020, and the Beacon Chain began producing blocks on 1 December 2020 at 12:00:23 UTC, after it had received the 16,384 deposits of 32 ETH it needed. For almost two years, it ran beside the mined chain without carrying any user transactions.
A new fee design: London and EIP-1559 (August 2021)
The whitepaper sent the whole fee to the miner. EIP-1559, created on 13 April 2019 by Vitalik Buterin, Eric Conner, Rick Dudley, Matthew Slipper, Ian Norden and Abdelhamid Bakhta, split it in two. Each block has a base fee that rises when blocks are full and falls when they are empty, and that base fee is burned, destroyed for good. The block producer keeps only an optional tip. Users get a more predictable price, and the network gets a link between usage and supply.
EIP-1559 went live with the London upgrade at block 12,965,000 on 5 August 2021. It changed the monetary debate around ether. The whitepaper’s plan was a fixed linear issuance. From London on, supply depends on how much is burned, so it can grow or shrink from one week to the next.
The Merge and the staking years (2022-2025)
On 15 September 2022, at block 15,537,394, the mined chain handed over to the Beacon Chain. Mining stopped, and validators who lock ether as collateral took over block production under proof of stake. ethereum.org’s roadmap says the switch reduced Ethereum’s energy consumption by about 99.95 percent. The mechanics of validators, slots and finality are covered in our in-depth lesson on proof of stake; a later lesson of this path, the-merge-and-ethereum-pos, tells the story of the switch itself. The forecast written in 2013 took nine years to come true.
The upgrades that followed each solved a specific problem. Shapella, at block 17,034,870 on 12 April 2023, let stakers withdraw ether, which had been locked since 2020. Dencun, at block 19,426,587 on 13 March 2024, introduced blobs through EIP-4844, a cheaper kind of data space built for rollups. Pectra, at block 22,431,084 on 7 May 2025, raised the maximum effective balance of a validator to 2,048 ETH, keeping the 32 ETH minimum. Fusaka, at block 23,935,694 on 3 December 2025, added PeerDAS, which lets nodes check rollup data by sampling it instead of downloading all of it.
Together, Dencun and Fusaka are the network’s answer to the whitepaper’s scaling worry quoted at the top of this lesson. The base chain stays small enough for ordinary machines, and the heavy traffic moves to rollups that pay for data space on it.
| Upgrade | Date (UTC) | Block or epoch | What it changed |
|---|---|---|---|
| Whitepaper | 27 November 2013 (ethereum.org timeline) | none | The proposal |
| Ether sale | 22 July to 2 September 2014 | none | About 31,000 BTC raised |
| Frontier | 30 July 2015 | Genesis | Network launch |
| Homestead | 14 March 2016 | 1,150,000 | Changes that eased later upgrades |
| DAO fork | 20 July 2016 | 1,920,000 | Funds of The DAO moved; Ethereum Classic splits off |
| Byzantium | 16 October 2017 | 4,370,000 | Protocol upgrade of the mining era |
| Beacon Chain | 1 December 2020 | Slot 0 | Proof of stake starts in parallel |
| London | 5 August 2021 | 12,965,000 | EIP-1559, base fee burned |
| Paris (The Merge) | 15 September 2022 | 15,537,394 | Proof of stake replaces mining |
| Shapella | 12 April 2023 | 17,034,870 (epoch 194,048) | Staking withdrawals |
| Dencun | 13 March 2024 | 19,426,587 (epoch 269,568) | Blobs for rollups (EIP-4844) |
| Pectra | 7 May 2025 | 22,431,084 (epoch 364,032) | Validators up to 2,048 ETH |
| Fusaka | 3 December 2025 | 23,935,694 (epoch 411,392) | PeerDAS data sampling |
| Glamsterdam | Not scheduled on mainnet (9 November 2026) | Sepolia testnet only, 6 October 2026 | Proposer-builder separation, block-level access lists |
The table gives the whole of Ethereum history in upgrades. Dates and blocks come from the ethereum.org history page; epochs are given from Shapella on, because upgrades after the Merge are switched on at a consensus-layer epoch rather than at a block height.
Ether and US regulators: from The DAO report to ETFs
Ethereum’s legal history in the United States is a series of separate steps, none of which settled every question. The first was the 2017 report on The DAO: tokens sold to fund a project could be securities. The second was a speech. On 14 June 2018, William Hinman, then director of the SEC’s Division of Corporation Finance, said in San Francisco that, based on the network’s decentralised structure, “current offers and sales of Ether are not securities transactions”. He set aside the 2014 fundraising, and the speech carries the usual note that it expresses his own views, not those of the Commission. It was a staff speech, not a rule.
The third step came in 2024. On 23 May 2024 the SEC approved rule changes letting NYSE Arca, Nasdaq and Cboe BZX list exchange-traded products holding ether, in an order published in the Federal Register (Release No. 34-100224). The funds began trading in July 2024. The listings are filed under exchange rules for “commodity-based trust shares”, the category used for commodity funds; the order is about listing rules and does not settle every question about ether.
For readers in the US and UK the tax side is simpler than the history. The IRS says that for US tax purposes “digital assets are considered property, not currency”, and brokers report sales on the new Form 1099-DA for transactions from 1 January 2025; details are in our guide to crypto taxes in the United States. In the UK, HMRC’s guidance says Capital Gains Tax may be due when you sell, exchange or give away cryptoassets, ether included; our UK crypto tax guide covers the rules.
Ethereum in November 2026: Glamsterdam and the next steps
The next upgrade is Glamsterdam, named after Amsterdam on the execution side and the star Gloas on the consensus side. Its two headline features are enshrined proposer-builder separation, which writes into the protocol the split between those who build blocks and the validators who propose them, and block-level access lists, which declare in advance what parts of the state a block touches so that clients can process it in parallel. It went live on the Sepolia testnet on 6 October 2026 at 13:53:36 UTC, as we reported at the time.
Since then the schedule has slipped. The activation on the Hoodi testnet planned for 26 October was postponed on 22 October, according to an Ethereum Foundation post that gave no new date. On 6 November another EF post explained that security firm ChainSecurity had found projects that would become vulnerable after two gas-pricing changes of the upgrade, EIP-8037 and EIP-8038, and that core developers had decided to delay the hard-fork process and not fork Hoodi until the issues were resolved. As of 9 November 2026 no mainnet date is set, even though the ethereum.org roadmap still labels Glamsterdam “Q4 2026”. The upgrade after it, Hegotá, is labelled 2027.
How the work is shared between the base chain and rollups after Dencun and Fusaka, and what Hegotá and later upgrades will add, is the subject of a later lesson of this path, ethereum-roadmap. For now the pattern is the same as in 2016: client teams agree on a set of changes, test them on public testnets, and only then pick a mainnet block or epoch.
Thirteen years of Ethereum history: what the whitepaper got right and what changed
Read today, much of the paper’s core design still holds. Accounts with code, gas as a meter for computation, contracts that hold money and call each other, tokens as a few lines of code, applications beyond payments, a move to proof of stake one day: all of these describe Ethereum in 2026. ethereum.org still keeps the document online because, in its words, it remains an accurate representation of the vision.
What changed is mostly the machinery around that core. The table sums it up.
| Topic | Whitepaper (2013-2014) | Ethereum in November 2026 |
|---|---|---|
| Consensus | Mining, with a move to proof of stake “likely” in the future | Proof of stake since 15 September 2022 |
| Mining puzzle | Contract execution as mining, called “untested” | No mining since the Merge |
| Fees | Whole fee paid to the miner | Base fee burned, tip to the block proposer (EIP-1559, August 2021) |
| Supply | Linear issuance with no cap, about 72 million ETH at launch (our computation) | Staking issuance minus burn; 122.2 million ETH on 9 November 2026 |
| Scaling | State trees and light clients, no rollups | Rollups using blobs (March 2024) and PeerDAS (December 2025) |
| Fork choice | Simplified GHOST with uncles | Proof-of-stake fork choice, no uncles |
| Uses | Sub-currencies, derivatives, names, storage, DAOs | Stablecoins, DeFi, NFTs, tokenised funds, DAOs |
Two changes stand out. The first is money: the paper sold a linear supply as a fairness device, while the network of 2026 burns part of every fee and pays stakers instead of miners. The second is scaling. The paper wanted one chain that stays light enough for ordinary nodes. Ethereum got there, but by pushing most activity onto other chains, a structure that its 2013 author did not describe and that is still being argued about.
How to read the Ethereum whitepaper today
Start at the version on ethereum.org, which opens with a warning that “the original whitepaper no longer reflects what Ethereum is today”, followed by a short list of what changed. Read that box first, then keep it in mind.
For a first reading, follow this order: the introduction and the four limits of Bitcoin scripting, “Ethereum Accounts”, “Messages and Transactions”, the “Applications” section, “Currency And Issuance” and the conclusion. That is most of the ideas in under an hour. Skip the code examples, written in Serpent, an early language that Solidity replaced, and the sections on Merkle trees and the modified GHOST protocol until later.
Then read with three questions. Is this still how it works? The account model and gas, yes. Did this happen? Tokens, DAOs and data feeds, yes; contract execution as mining, no. What is missing? Rollups, staking, the fee burn and every token standard. For the protocol as it runs now, the Yellow Paper is the formal reference, and the ethereum.org roadmap lists what comes next. If you want the Bitcoin counterpart, read our guide to Satoshi Nakamoto’s Bitcoin whitepaper. In this path, the next lessons, how-the-evm-works and the-merge-and-ethereum-pos, go into the machine and the switch to staking.
