If you have not checked in on Web3 since 2021 you would probably not recognise it. The noise is quieter. The pitches are duller. And the people actually building are harder to distinguish from the people running institutions than they used to be.
Web3 in 2026 is something narrower and more durable than the decentralised consumer revolution that was marketed during the last bull run. It is a set of infrastructure technologies that institutions, governments, and large platforms are absorbing into regulated financial systems. The open networks, programmable money, and censorship-resistant settlement that the original pitch was built on are still there. They have mostly been embedded inside compliance frameworks rather than replacing them.
What actually changed
Two things in 2025 permanently rearranged the landscape. First, President Trump signed the GENIUS Act into law in July 2025 with bipartisan support: the Senate passed it 68 to 30, the House 308 to 122. It established the first federal regulatory framework for payment stablecoins in the US, requiring 100% reserve backing with liquid assets like US dollars or short-term Treasuries, monthly public disclosures of reserve composition, and a dual regulatory regime where federally chartered banks and non-bank issuers can apply for licenses. Algorithmic stablecoins were effectively prohibited. As Paul Simroth reported in his May 2026 outlook, stablecoins issued under GENIUS are “explicitly carved out of the definitions of security and commodity under federal securities laws and the Commodity Exchange Act, removing them from SEC and CFTC jurisdiction.”
Second, the EU’s Markets in Crypto-Assets regulation (MiCA) had been in full application since December 2024. More than EUR 540 million in fines had been issued and at least 63 firms had their licenses revoked by late 2025 for AML, KYC, or reserve failures. The patchwork of permissive and hostile jurisdictions that defined crypto regulation for a decade gave way to a coordinated global framework.
The downstream effect was immediate. USD-backed stablecoins grew from $128 billion to $225 billion in combined supply between 2024 and April 2025, a 76% increase. By spring 2026 the total stablecoin market capitalisation sat near $300 billion.
The things that are actually working
Real-world asset tokenisation
This barely existed as a category in 2023. By April 2026, tokenised real-world assets had reached approximately $27.6 billion in on-chain value, growing from around $5 billion at the start of 2025. Tokenised US Treasuries formed the largest category at roughly $12.88 billion, with BlackRock’s BUIDL fund as the single largest product. JP Morgan’s Kinexys blockchain platform has processed over $1.5 trillion in cumulative transaction volume since its 2019 launch, with average daily volume exceeding $2 billion. Ondo Finance launched tokenised US stocks and ETFs on Ethereum, BNB Chain, and Solana throughout 2025 and into 2026. Coinbase announced a Tokenize platform for full tokenised equities. Robinhood introduced over 200 tokenised stock and ETF tokens for European customers on Arbitrum.
The IMF published a paper in April 2026 titled “Tokenized Finance” authored by Tobias Adrian that argues tokenisation “represents a structural shift in financial architecture rather than a marginal efficiency gain.” The language matters primarily because it came from the IMF. McKinsey estimates the market could reach $2 trillion by 2030. Standard Chartered forecasts $30 trillion by 2034. The gap between current reality and those projections tells you where the execution risk sits.
Layer 2 consolidation
Layer 2 networks now account for roughly 95% of Ethereum’s total transaction throughput. Total value locked grew from under $4 billion in 2023 to approximately $47 billion by early 2026. Ethereum mainnet gas fees fell from around 7 gwei in January 2025 to roughly 0.43 to 0.50 gwei in January 2026, a 93% drop. L2 transactions routinely cost well below one cent.
The concentration pattern is stark. As Simroth put it: “Most new L2s launched in 2025 became ghost towns after their airdrop farming cycles ended.” Base, Arbitrum, and Optimism now account for roughly 90% of all L2 transactions. Enterprise adoption of the rollup model accelerated: Kraken launched INK, Uniswap launched UniChain, Sony launched Soneium for gaming and media distribution, and Celo and Lisk transitioned from Layer 1s to OP Stack L2s. “Raw TPS numbers are no longer the differentiator,” Simroth wrote. “Ecosystem gravity is.”
AI agents with crypto wallets
This is the trend that feels most like 2021 hype rebranded, and some of it is. But a narrow slice is real. Autonomous AI agents operating their own crypto wallets that transact on-chain started to generate measurable volume on Base and Solana in early 2026. Coinbase launched Agentic Wallets in February 2026 specifically for this use case, with non-custodial wallets secured in Trusted Execution Environments.
The x402 payment standard, which Coinbase published in May 2025 with a V2 release in December 2025, had processed over 100 million payments by early 2026 and is now governed by the x402 Foundation co-founded with Cloudflare and including Google, Visa, AWS, Circle, and Anthropic as members. Simroth described the reasoning: “Agents operate globally, programmatically, and 24/7, which is exactly what stablecoins and smart contracts are good at and which traditional banking infrastructure is poor at.” The difference between this and the 2021 crypto-gaming narrative is that this time there are actual transactions to measure.
On-chain data as infrastructure
Dune now indexes over 100 blockchains with 1.5 million datasets and 6.5 million queries. Its AI agents product launched in 2026. Nansen provides real-time wallet labelling and smart money flows. These platforms have become the reference layer for how institutions, analysts, and AI systems understand on-chain activity. When the IMF writes about tokenised finance, the data it cites runs through these pipes. When an AI visibility audit checks whether a protocol’s on-chain activity matches its public narrative, Dune is where the evidence lives. The existence of a mature data infrastructure layer is itself a signal: it means the worst of the informational chaos that defined early Web3 has passed.
The things that did not happen
Consumer Web3 mostly did not arrive. The gaming pivot, the metaverse push, the idea that normal people would use blockchain apps without knowing or caring about the underlying tech. That vision contracted hard. Retail attention is lower than it was in 2021. The projects that survived on that thesis either pivoted to infrastructure or faded.
The decentralised autonomous organisation (DAO) model as originally pitched, where thousands of token holders govern a protocol through on-chain voting, remains largely unworkable for anything requiring speed or nuance. Most DAOs have either centralised into de facto core teams or become ceremonial.
Permissionless DeFi, the idea that anyone could lend, borrow, or trade without identity verification, is under structural pressure. MiCA’s enforcement actions and the FATF’s Travel Rule, now adopted by 85 of 117 surveyed jurisdictions covering approximately 98% of the global virtual asset market, have made KYC-compliant rails the default path. Simroth framed it honestly: “Whether this creates a healthy bifurcation or a chilling effect on innovation is the central political question of the 2026 crypto cycle.”
What this means for the original premise
The original Web3 vision was about user control: you own your data, your identity, your assets, and no platform can take them away. That vision has not died. It has been absorbed. The EU’s Digital Identity Framework requires all 27 member states to provide compliant digital wallets by December 2026, using the same cryptographic primitives that self-sovereign identity was built on. The infrastructure is being deployed by governments rather than startups, which complicates the ideological narrative but validates the technical one.
The programmable money thesis has been validated more clearly than anyone expected. Stablecoins are now a $300 billion regulated asset class. Tokenised Treasuries exist at institutional scale. The question is no longer whether blockchain-based financial infrastructure works. It is whose regulatory framework it operates under and who controls the settlement layer.
Frequently asked questions
Is Web3 dead in 2026?
No, but the consumer vision of Web3 that peaked in 2021 has largely not materialised. The institutional and infrastructure side, regulated stablecoins, RWA tokenisation, L2 scaling, on-chain data platforms, has grown significantly. What is happening now looks more like fintech infrastructure built on blockchain rails than the decentralised consumer internet that was originally pitched.
What happened to NFTs and the metaverse?
The speculative NFT market of 2021 collapsed and has not recovered at the same scale. Utility-based NFTs exist in specific niches like ticketing, loyalty programmes, and gaming assets. The metaverse as a consumer destination remains unrealised. Projects that built on that thesis have pivoted or faded.
Is DeFi still a thing?
DeFi protocols continue to operate and process significant volume. The permissionless, anonymous version of DeFi is under regulatory pressure and has mostly relocated offshore. The regulated, KYC-compliant version is growing within the stablecoin and RWA frameworks.
What should a builder build on in 2026?
Base, Arbitrum, and Solana have the most ecosystem gravity. The technical choice of rollup type matters less than the distribution and liquidity you get access to. On-chain data infrastructure (Dune, Nansen) has matured enough to serve as a reliable reference layer. The x402 framework for AI agent payments is the most interesting new rails for programmable transactions.
How does AI visibility fit into this?
Most Web3 projects in 2026 still have fragmented, outdated, or inconsistent public information. As AI systems become the default entry point for research, the gap between what a project actually is and what models say it is widens. The AI visibility audit checks this.
Leave a Reply