Tiger Research: How Crypto Giants Are Betting on AI Agent Payment Infrastructure

This report is authored by Tiger Research. To achieve true autonomy, native payment capability is essential. The market has already begun actively positioning for this shift.

Core Points

• The payer is shifting from humans to AI Agents, making payment infrastructure a core requirement for achieving true autonomy.
• Big Tech companies (including Google AP2 and OpenAI Delegated Payment) are designing approval-based automated payment systems on top of existing platform infrastructure.
• Cryptocurrency enables a disintermediated payment model through standards like ERC-8004 and x402, utilizing NFT-based identity and smart contracts.
• Big Tech prioritizes convenience and consumer protection, while crypto emphasizes user sovereignty and broader Agent-level execution capabilities.
• The key future question is: Will payments be controlled by platforms or executed by open protocols?

1. Payments Are No Longer Exclusive to Humans

Source: macstories (Feder1C0 Viticci)

Recently, "OpenClaw" has garnered significant attention. Unlike AI systems like ChatGPT or Gemini, which primarily focus on retrieving and organizing information, OpenClaw allows AI Agents to execute tasks directly on a user's local PC or server.

Through instant messaging platforms like WhatsApp, Telegram, and Slack, users can issue instructions, and the Agent then autonomously executes tasks, including email management, calendar coordination, and web browsing.

As it runs as open-source software and is not tied to a specific platform, OpenClaw functions more like a personal AI assistant. This architecture is favored for its flexibility and user-level control.

However, a critical limitation remains. For AI Agents to achieve full autonomy, they must be able to execute payments. Currently, Agents can search for products, compare options, and add items to a cart, but final payment authorization still requires human approval.

Historically, payment systems were designed around human actors. In an AI Agent-driven environment, this assumption no longer holds. If automation is to become fully autonomous, Agents must be able to independently evaluate, authorize, and complete transactions within defined constraints.

Anticipating this shift, both Big Tech companies and crypto-native projects have introduced technical frameworks over the past year aimed at enabling Agent-level payments.

2. Big Tech: Building Agent Payments on Existing Infrastructure

In January 2025, Google launched AP2 (Agent Payment Protocol 2.0), expanding its AI Agent payment infrastructure. While OpenAI and Amazon have also outlined related initiatives, Google is currently the only major player with a structured implementation framework.

AP2 divides the transaction process into three Mandate Layers. This structure allows for independent monitoring and auditing of each stage.

• Intent Mandate: Records the action the user wants to perform.
• Cart Mandate: Defines how the purchase should be executed under preset rules.
• Payment Mandate: Executes the actual transfer of funds.

Example: Suppose Ekko instructs an AI Agent on Google Shopping to "find and buy a winter jacket under $200."

• Intent Mandate: Ekko instructs the AI Agent to purchase "a winter jacket with a maximum budget of $200." This information is recorded on-chain as a digital contract, the Intent Mandate.
• Cart Mandate: The AI Agent follows the intent, searches for matches among partner merchants, and adds eligible items to the cart. Verifies price ($199, within budget ✓), confirms shipping address.
• Payment Mandate: Ekko reviews the selected item and clicks approve. The $199 is processed via Google Pay. Alternatively, the AI Agent can automatically complete the payment within preset parameters.

Throughout this process, the user does not need to input additional information. Google AP2 relies on existing user credentials (pre-registered cards and addresses), which lowers the barrier to entry and simplifies adoption.

Source: Google

However, Google currently only supports Agent payments for companies within its partner network. Therefore, its usage is confined to a controlled ecosystem, limiting broader interoperability and open access.

3. Cryptocurrency: Self-Custody and Open Exchange

The crypto space is also developing payment infrastructure for AI Agents, but its approach is fundamentally different from Big Tech's. While big platforms build trust within controlled ecosystems, crypto starts from a different question: Can AI Agents be trusted without relying on centralized platforms?

Two core standards aim to address this goal: Ethereum's ERC-8004 and Coinbase's x402.

First is the identity layer. AI Agents operating on the blockchain must be identifiable. ERC-8004 serves this function. It is issued as an NFT, but not as an art collectible; it is a credential NFT containing structured identity data. Each token consists of three parts:

1. Identity
2. Reputation
3. Validation

These elements together form a verifiable on-chain identity certificate.

Regarding the payment mechanism, x402 serves as the payment pathway. Developed by Coinbase, x402 is a crypto-native payment standard for AI Agents. It enables Agents to conduct autonomous transactions using stablecoins. Its core feature is automated smart contract execution, where conditional logic is embedded directly into the code, allowing settlement to occur without human intervention once conditions are met.

When ERC-8004 (identity) is combined with x402 (payment), AI Agents can verify counterparties and execute transactions without relying on a centralized platform.

Example: Ekko instructs his Agent A to buy a used laptop with a maximum budget of $800. The seller's Agent B communicates directly with it.

• Mutual Verification: Check identity and reputation score via ERC-8004 NFT (e.g., Reputation 72, balance confirmed).
• Smart Contract Escrow: $800 is transferred from the wallet into a smart contract escrow, locking the funds until delivery is confirmed.
• Settlement and Reputation Update: After the transaction is completed, x402 automatically settles, and the reputation records of both parties are automatically updated and written into their respective ERC-8004 NFTs.

Throughout this process, no intermediaries are involved. The two AI Agents transact directly through blockchain-based verification and settlement, embodying the crypto-native model of Agent-to-Agent (A2A) commerce.

4. Big Tech vs. Cryptocurrency: Differences in the AI Agent Operational Domain

Google AP2 represents a controlled model designed for verified partners. Google restricts market participants to protect consumers. Since AI Agent execution has probabilistic outcomes rather than being fully deterministic, liability for transaction errors may ultimately fall on the payment infrastructure provider. To reduce the probability of failure, Google is incentivized to narrow its ecosystem.

A restricted ecosystem increases stability but also limits the Agent's ability to operate autonomously and optimize choices across a broader market.

In contrast, ERC-8004 and x402 reflect a more open architecture. The crypto model aims for permissionlessness and interoperability.

While end-to-end execution is not yet perfect, the long-term vision is for Agents to independently manage daily consumption. Big platforms may attempt to integrate major retail channels, while open crypto standards have a structural advantage in handling small, high-frequency, programmatic payments (micropayments). For example, if an Agent buys 1000 stock images at $0.01 each, the crypto-native path offers higher operational efficiency.

Of course, the lack of a central authority involves trade-offs: identity evaluation standards must be established in a decentralized manner, and no single entity bears ultimate responsibility for failure.

Conclusion

Both Big Tech and the crypto space are pursuing the same goal: enabling autonomous AI Agent commerce. The difference lies in the architecture: Big Tech favors closed, controlled systems, while crypto advances open, protocol-based models.

The future trend is more likely to involve interoperability between the two approaches, rather than a zero-sum game.