'Web3 Will Replace the Internet': What Happened

On 21 December 2021, Jack Dorsey, who had stepped down as Twitter's chief executive a month earlier, posted a short message about the hottest idea in technology. "You don't own 'web3'," he wrote. "The VCs and their LPs do." Elon Musk joked the same week that he could not find web3 anywhere. Dorsey replied that it was "somewhere between a and z", a jab widely read as aimed at Andreessen Horowitz, the venture firm known as a16z. The next day he added that "the VCs are the problem, not the people". Marc Andreessen, a16z's co-founder, blocked him on Twitter, and the exchange became one of the best-known arguments of the Web3 boom.

The jab stung because of the money. Venture capital firms, which invest other people's money in young companies for a share of the profits, put about $32.8 billion into crypto start-ups in 2021 and more than $30 billion in 2022, according to Galaxy Digital's research. The 2021 figure alone was more than all previous years combined. Much of it went to companies promising "Web3".

The promise

The pitch divided the internet into three ages. Web1 was the read-only web of the 1990s. Web2 was the web of Google, Facebook and Twitter, where anyone could post but a few companies owned the platforms, the data and the advertising money. Web3 would hand ownership back to users.

The tool was the blockchain, a shared ledger kept by thousands of computers that no single company controls. Apps would run on smart contracts, pieces of code that move money or record ownership by themselves, like a vending machine that needs no shopkeeper. Users would hold tokens, digital shares in the services they used, and vote on how those services were run. A social network would belong to its members, not to Mark Zuckerberg.

a16z raised a $2.2 billion crypto fund in June 2021 and a $4.5 billion one in May 2022. Its partner Chris Dixon became the movement's best-known advocate. The phrase "Web 3.0" had been coined in 2014 by Gavin Wood, a co-founder of Ethereum, the blockchain most Web3 apps were built on.

The problem

Dorsey's point was about who owned what. Venture funds raise money from "limited partners", or LPs, such as pension funds and university endowments, who expect big returns. In many Web3 projects, the insiders and funds held a large share of the tokens from the start. Ownership was spread out on paper and concentrated in practice, a little like a company that lists its shares on the stock market but keeps most of them with the founders and early backers.

The products had problems too. In January 2022 Moxie Marlinspike, the founder of the encrypted messaging app Signal, wrote an essay about building his own Web3 apps. He found that most of them did not talk to a blockchain directly at all. They relied on a handful of companies such as Infura, Alchemy and the NFT marketplace OpenSea to do it for them. The new internet, he argued, was being rebuilt around new middlemen.

Ordinary users met slow apps, fees on every action, confusing digital wallets, and passwords that could not be reset: lose the "seed phrase" that unlocks a wallet and the money inside is gone for good. A free app from Google or Meta asked for none of that. Few people switched.

The crash and the retreat

Then 2022 happened. The collapse of Terra in May, the lender Celsius in June and the exchange FTX in November erased about two-thirds of the crypto market's value and most of the public's patience. NFT sales, once Web3's showpiece, fell away. "Web3" disappeared from start-up pitch decks, replaced by artificial intelligence.

Dorsey's own alternative did not last either. His payments company, Block, launched a unit called TBD in 2021 to build what it called "Web5", a decentralised web built on bitcoin. Block announced in November 2024 that it was winding TBD down to concentrate on bitcoin mining and its Bitkey wallet.

The clearest test case was Farcaster, a decentralised social network often described as the Web3 answer to Twitter. In 2024 its developer, Merkle Manufactory, raised $150 million at a valuation of about $1 billion. In December 2025 co-founder Dan Romero said it had about 250,000 monthly users; earlier analyses found that many of its daily users were bots. In January 2026 the protocol was handed to Neynar, a company that built tools for it, the founders stepped back, and Merkle said it would return all $180 million it had raised to its investors. Romero insisted Farcaster was "not shutting down". Facebook, Instagram and WhatsApp, by comparison, are each used by billions of people.

What survived

Some of the plumbing proved useful, just not as a replacement for the internet. Stablecoins, crypto tokens pegged to the US dollar, are the clearest success: their total value was about $300 billion in mid-2026, according to DefiLlama, and the US passed a law to regulate them, the GENIUS Act, in July 2025. People use them to move dollars across borders and to trade. Decentralised finance, known as DeFi, where lending and trading run through smart contracts instead of banks, held about $100 billion in March 2026, below its 2021 peak.

None of this rebuilt the web. Big technology companies kept their users, and banks and payment firms began adopting stablecoins on their own terms. The internet absorbed the parts of crypto that worked and left the rest.