Web4.0 Track: The Real Stage in 2026

Since 2026, a concept deeply related to AI has frequently appeared in crypto research and investment institutions: Web4.0. Unlike the previously popular narratives, this time the focus of discussion is not on new consensus mechanisms or higher throughput, but on a switch in the executing主体—from humans operating the internet to autonomous AI Agents directly participating in economic activities.

The GoToMars Research Institute has been continuously tracking the integration of AI and crypto. This report begins with progress in infrastructure development, assessing the current stage of this track as of April 2026, including structural opportunities and practical constraints.

1. Long-neglected structural mismatches

Before discussing Agent economies, it is necessary to understand a real-world problem: current AI models already possess considerable perception and decision-making capabilities, but lack almost any authority to act in business contexts. Every link in the traditional financial system—from account opening, identity verification, to payment authorization and contract signing—is built on the basis of real-name registration of individuals or legal entities. An AI program running in the cloud cannot legally open a commercial bank account nor independently sign legally binding agreements. As a result, AI can only serve as a tool to assist humans, making it difficult to assume the role of an independent economic主体.

A deeper mismatch appears in payment granularity. Taking Stripe’s standard fee rate as an example, a single payment incurs a fixed cost of 2.9% plus $0.30. This pricing assumes larger, less frequent human consumption behaviors. But the actual behavior pattern of Agents is the opposite: a single API call might only cost $0.001 to $0.01, yet the call frequency can reach dozens per second. Under such transaction characteristics, the fixed fees of traditional card organizations exceed the transaction amount itself, making micro-payments unfeasible. The table below summarizes the differences between two user types across several key dimensions.

Table 1: Key Interaction Feature Differences Between Human Users and AI Agents

This mismatch means that, even with strong model capabilities, AI can only play the role of an information processor within the current financial system. The prerequisite for turning Agents into economic主体 is to find a set of underlying protocols that break out of the traditional financial identity framework.

2. Why encrypted infrastructure becomes an alternative path

Permissionless blockchain networks happen to provide a structural solution to the above issues. On networks like Ethereum, Solana, or TRON, generating an on-chain address only requires locally creating a public-private key pair once, a process completed in milliseconds without any centralized approval. This means AI Agents can create independent identities for themselves or their derived sub-tasks at any time and directly participate in on-chain transactions as asset承接方. The introduction of stablecoins further solves the value unit problem: USDT and USDC currently have a combined circulation exceeding $250 billion, providing stable pricing and settlement media. Coupled with smart contracts and composable DeFi protocols, complex financial operations such as fund management, lending, and hedging can be performed via programmatic interfaces, no longer limited by business hours or manual approval.

This alternative path is precisely the core reason why recent institutional investors are optimistic about Crypto in the Agent economy—its core logic lies in the fact that: the deterministic and verifiable nature of crypto systems better matches machine behavior needs—AI will not fatigue or forget, and can complete contract audits and address verification within seconds, which has long been a weak point for human users. Under this narrative, Web 4.0 is born, defined as:

Web 4.0 = Web 3.0 + AI Agent

The underlying proposition of Web4.0 is thus established: combining encrypted infrastructure with autonomous AI Agents, enabling machines to move from the information layer to the economic layer.

3. Progress and constraints of three key protocols

For Agents to truly become economic主体, three protocol gaps must be filled on top of existing blockchain infrastructure: payments, identity, and tool invocation. Over the past twelve months, these gaps have been filled by three representative protocols, all making concentrated progress from late 2025 to early 2026.

Figure 2: Timeline of Key Milestones in Web4.0 Infrastructure Development(May 2025 to April 2026)

The payment layer was filled by Coinbase’s x402 protocol, released in May 2025, which reuses the long-idle 402 Payment Required status code in HTTP, allowing servers to directly return a payment request in a single HTTP response. Clients can complete payment by signing with stablecoins and then access resources. On April 2, 2026, the x402 Foundation officially joined the Linux Foundation for open-source governance. The foundation was initiated by Coinbase, Cloudflare, and Stripe, with governance involving over 20 organizations including AWS, American Express, Google, Mastercard, Microsoft, Polygon Labs, Shopify, Solana Foundation, and Visa.

The identity layer was filled by ERC-8004, led by the Ethereum ecosystem. Drafted jointly by MetaMask, Ethereum Foundation, Google, and Coinbase representatives, it completed reference registry deployment on the Ethereum mainnet on January 29, 2026.

The tool invocation layer was led by Anthropic’s MCP protocol, open-sourced in November 2024, and donated to the newly established Agentic AI Foundation under the Linux Foundation on December 9, 2025.

It is noteworthy that the pace of protocol development significantly outpaces actual commercial traffic accumulation. According to Unchained’s report on April 3, 2026, the x402 protocol on the Base network handles about 55k transactions daily, totaling approximately 97 million transactions on Base. Artemis, an on-chain analytics firm, reports that the cross-chain daily transaction volume of x402 is about $28k, with an average transaction of roughly $0.20. Meanwhile, about half of these transactions are estimated to be spam or testing, with real commercial traffic still in the very early stage. Therefore, the time lag between protocol standardization and commercial adoption is the most critical observation point at this stage.

Table 2: Progress and Practical Constraints of the Three Key Protocols in Web4.0

Sources: Linux Foundation, Ethereum Foundation, Anthropic, The Block, Unchained disclosures; compiled by GoToMars Research Institute. Data as of mid-April 2026.

4. Typical case: computing power layer—early recognized as a foundational asset by capital markets

In the early stages of payment protocols and application scenarios, the most bottom-layer computing assets in the entire Web4.0 stack have been the first to connect with traditional financial channels. This is an easily overlooked but highly signaling phenomenon—traditional capital did not wait for Agent applications to be fully operational before entering, but chose to establish exposure at the computing power layer first.

The most典型 sign is Grayscale. On December 30, 2025, it submitted a Form S-1 registration statement to the SEC to convert the Grayscale Bittensor Trust into a spot ETP, aiming to list on NYSE Arca under the ticker GTAO. On the same day, Bitwise also submitted registration applications for eleven strategy-based crypto ETFs (Form N-1A), including one tracking TAO—the Bitwise TAO Strategy ETF. On April 2, 2026, Grayscale further submitted an Amendment No. 1 to the S-1, pushing the product closer to official listing. In Switzerland, Deutsche Digital Assets and Safello jointly launched the Safello Bittensor Staked TAO ETP (ticker: STAO) on November 19, 2025, traded in USD, fully backed by physical assets, with staking yields. This means that even before North American ETFs are approved, Europe has already introduced TAO-related products into mainstream exchanges.

Institutional secondary asset allocation is also accelerating. On April 7, 2026, Grayscale rebalanced its Decentralized AI Fund, increasing TAO’s weight from 31.35% to 43.06%, the largest single-asset weight change in this adjustment, without adding or removing other assets. This weight shift indicates that institutions are beginning to incorporate decentralized AI into the main AI narrative—previous institutional expressions of this theme were almost entirely through NVIDIA and a few large model vendors; TAO is the first crypto-native asset to be significantly reallocated by mainstream decentralized AI thematic funds.

The scarcity on the supply side further reinforces this long-term trend. Bittensor’s halving on December 14, 2025, reduced daily issuance from 7,200 to 3,600 TAO, following Bitcoin’s supply rhythm (Grayscale Research, December 2025; Bittensor official documentation). With about 70% of circulating supply staked long-term, limited remaining supply absorbed by listed companies’ treasuries and ETP structures, the liquidity pressure of computing power assets is structurally tightening. This is the fundamental reason why institutions are willing to establish positions at the computing layer before application deployment.

However, a real risk must also be noted. On April 10, 2026, Covenant AI, the operator of the leading subnet of Bittensor, announced its withdrawal from the network and sold approximately 37k TAO, causing a temporary 20%+ price correction and raising concerns about network centralization. This event reminds institutional funds that, although the logic of assetizing computing power is valid, the governance structure and network stability are still in a period of dual adjustment, and cannot be simply compared to early Bitcoin institutional acceptance.

The signals from the computing layer have two implications. First, institutional channels opening means node operators, staking service providers, subnet developers, and compliant custodians built around this layer will benefit from structural liquidity inflows—this is a direction for deep participation in the next stage. Second, when institutions link TAO with decentralized AI as a configurable asset, valuation anchors for the entire Web4.0 stack will gradually establish, and the assetization paths of the payment, identity, and application layers will gain reference points.

5. GoToMars Research Institute’s key judgments

Based on the above observations, the Web4.0 track is currently in a stage of infrastructure concentration and relatively lagging commercial adoption. Narratives, capital, and protocols have initially converged, but real on-chain commercial traffic remains in early validation. For investment institutions and project teams, this stage presents both opportunities and risks—success depends on accurately judging the mismatch in timing between protocol windows and commercial windows.

Table 4: GoToMars Research Institute’s three key judgments on the Web4.0 track

Sources: GoToMars Research Institute, April 2026.

The first judgment is that the outcome of the standardization game will be clear within the next twelve months. Although x402 and ERC-8004 have achieved phased breakthroughs, real traffic formation still takes time. Once standards are established, toolchains, SDKs, and auditing services built around them will see concentrated opportunities. The second judgment concerns product migration: modular, programmable product architectures represented by Skills will gradually shift from AI trading to traditional industries such as insurance actuarial, supply chain fulfillment, and content copyright settlement. This is a key vertical incubation opportunity for the next phase. The third judgment is that competition in the stablecoin settlement layer will become the critical venue for actual payment layer deployment. Projects building comprehensive Agent financial operation systems around different stablecoin networks—such as TRON, Solana, Base, BNB Chain—have strong structural opportunities.

The window for bottom-layer infrastructure positioning is narrowing, while the application layer capable of transforming protocol capabilities into real commercial closed loops will be a more valuable opportunity between 2026 and 2027. The completion of the protocol layer determines the potential boundaries of upper-layer applications, but true value realization ultimately depends on service capabilities for machine users in specific scenarios.