Web2 vs web3 is a difference in who runs the backend and who holds the account. Web2 is the internet most people use today: apps run on servers owned by companies, and users sign in to accounts those companies control. Web3, in Ethereum's definition, refers to decentralized apps that run on a blockchain and let anyone participate without monetizing their personal data, per Ethereum's developer docs. Ethereum co-founder Gavin Wood coined the term "Web 3.0" in 2014, per Ethereum's introduction to Web3.
​

This page compares the two models on what changes in practice: logins, payments, data ownership, uptime, speed and cost. It covers both the user view and the developer view, plus the limits that keep most apps on web2 infrastructure today. Background: what a blockchain is and what a smart contract is.
​

Web2 apps store data and run logic on servers a company controls, so the company sets the rules, holds the user's account and can change or close it. Web3 apps put core logic in smart contracts on a public blockchain and identify users by wallet addresses they control, so no single operator can block access or rewrite balances.
​

Ethereum's introduction frames the web in three loose periods. Web 1.0, roughly 1990 to 2004, was mostly static company websites, so it is known as the read-only web. Web 2.0 began in 2004 with social media platforms, turning the web read-write, and it also brought the advertising-driven revenue model in which users create content but do not own it or share in its monetization, per Ethereum's introduction to Web3. Web3 is pitched as read-write-own.
​

The same source lists four core ideas for web3: it is decentralized, permissionless, has native payments in cryptocurrency, and is trustless, relying on incentives and economic mechanisms instead of trusted third parties, per Ethereum's introduction to Web3. These are design goals. How fully any given app meets them varies.

A web2 app is a client talking to a company's servers. The company stores user data in its own databases, runs the business logic, decides who gets an account, and processes payments through banks or card networks. This model is fast, cheap at scale and easy to update, because one operator controls every piece.
​

Ethereum's documentation credits centralization with onboarding billions of people and building the stable infrastructure the web runs on, while noting that a handful of centralized entities now decide what is and is not allowed on large parts of it, per Ethereum's introduction to Web3. Its developer comparison summarizes the general pattern: centralized systems usually have higher throughput, resolve conflicting data easily because the central authority is the source of truth, and can push upgrades with little friction, per Ethereum's developer docs.
​

The costs sit on the user side. A central authority is a single point of failure, it can censor data, and it controls who participates, per the same developer docs. Ethereum's examples are concrete: a social platform can remove any account or post, a payment service can refuse payments for certain types of work, and gig-economy servers can go down and interrupt worker income.
​

A web3 app, often called a dapp, keeps its core state and rules in smart contracts on a blockchain. The user connects a wallet, the app's interface builds a transaction, the user signs it with a private key, and the network executes it. Anyone can read the contract's state, and no operator can change a balance the contract holds.
​

Ethereum lists the reasons developers build dapps: anyone on the network can use the service without permission, no one can block or deny access, payments are built in through ETH, and Ethereum is Turing-complete, so developers can program nearly anything, per Ethereum's developer docs. More on open access: what a permissionless blockchain is.
​

Identity changes too. In web2, each platform needs its own account, and social sign-ins can lock a user out of many services at once. In web3, an Ethereum address and an Ethereum Name Service profile act as a single login across platforms, per Ethereum's introduction to Web3. The wallet that holds that address is covered in what a blockchain wallet is, and the security choices behind it in hot wallet vs cold wallet.
​

Most dapps are hybrids. Ethereum notes that successful dapps put only small portions of their code onchain because it is expensive, and that the web3 ecosystem still depends mainly on centralized infrastructure such as GitHub, Twitter and Discord, per Ethereum's developer docs and its introduction to Web3.
​

For users, web3 changes three things: ownership, portability and payments. Assets bought in a web3 app sit at the user's address instead of inside a company's account database. Reputation and data recorded onchain can move to another interface. Money moves as tokens without a bank in the path. In exchange, the user takes on key management.
​

Ethereum's example of ownership is a game item. In a web2 game, a purchased item is tied to an account, and if the developer deletes the account the item is gone. In web3, the item can be an NFT that no one, including the game's creators, can take away, and the player can sell it on an open market, per Ethereum's introduction to Web3.
​

Payments follow the same logic. Web2 payment infrastructure relies on banks and processors and can exclude people without bank accounts, while web3 uses tokens like ETH to send money directly in the browser, per Ethereum's introduction to Web3. Dollar-denominated stablecoins apply the same model to everyday money, covered in programmable money, and lending and trading apps built this way are covered in what DeFi is.
​

The trade-off is responsibility. A forgotten web2 password has a reset link. A lost web3 recovery phrase usually has no fix, and transactions cannot be reversed, per Ethereum's wallet guide.
​

For developers, the backend becomes a shared public network instead of a private server. Contract code and state are visible to anyone, every state change costs a transaction fee, and deployed logic is hard to change. Developers gain built-in payments, open data and composability with other contracts, and give up cheap iteration and the ability to quietly patch production.
​

Ethereum's developer comparison lists the limits a team will hit first. Scalability: state changes such as payments must be processed by a node and propagated across the network, so web3 transactions are slower. UX: users need extra steps, software and education. Accessibility: modern browsers lack native integration. Cost: onchain code is expensive, per Ethereum's developer docs.
​

The architecture choices that follow are mostly about what goes onchain. A common split keeps balances, ownership and settlement rules in contracts, and keeps search, media and analytics on conventional servers. Fees push many apps onto layer 2 networks; Ethereum's own introduction says broader layer 2 adoption is needed to make web3 affordable for users in developing economies, per Ethereum's introduction to Web3. Public state also means public transactions, which anyone can inspect through a blockchain explorer.
​

The web2 vs web3 split maps onto a broader trade-off between centralized and decentralized networks. Centralized systems are faster, simpler and easier to coordinate, but have a single point of failure and a gatekeeper. Decentralized systems resist censorship and outages, but run slower, cost more, and struggle to coordinate upgrades.
​

Ethereum's developer documentation sets out the trade-off directly. The excerpt below is quoted from its centralization table, per Ethereum's developer docs:
​

"Single point of failure: malicious actors may be able to take down the network by targeting the central authority." versus "No single point of failure: network can still function even if a large proportion of participants are attacked/taken out." (Ethereum developer docs)
​

The same page adds a caveat worth keeping: these are general patterns, and in reality every network sits somewhere on a spectrum, with none fully centralized or fully decentralized, per Ethereum's developer docs. A blockchain's consensus design is one place that spectrum shows up, covered in proof of stake vs proof of work.
​

Web3's limits are practical, and even its advocates list them. Transactions settle more slowly, setup takes more steps than a web2 sign-up, a lost key cannot be reset, and much of the ecosystem still runs on centralized tools. Most services do not need to become dapps, and Ethereum's own documentation says so.

Ethereum co-founder Gavin Wood coined the term "Web 3.0" in 2014, per Ethereum's introduction to Web3.
​

No. A blockchain is the infrastructure; web3 describes apps and services built on it, using blockchains, cryptocurrencies and NFTs to give users ownership, per Ethereum's introduction to Web3.
​

For most web3 apps, yes. A wallet lets you sign in to applications, read your balance and send transactions, per Ethereum's wallet guide.
​

State changes must be processed by a node and propagated across the whole network rather than written to one company database, per Ethereum's developer docs.
​

Web 1.0, roughly 1990 to 2004, was mainly static websites owned by companies with little user interaction, known as the read-only web, per Ethereum's introduction to Web3.
​

Definitions, periods and trade-offs come from the Ethereum Foundation's ethereum.org pages, which present web3 from Ethereum's perspective; competing definitions of web3 exist, and this page attributes each claim to its source rather than adopting the framing. All pages were read on October 13, 2026.