The agentic payment protocol x402 experienced a severe divergence between transaction volume and count between late 2025 and May 2026. Total value transferred plummeted 77% from a November 2025 peak of $5.15 million to $1.19 million, yet the number of transactions only declined 41% from a December 2025 high of 4.85 million before surging to 2.89 million in May. This rebound represents a 12.5x increase from the February trough, driven by an average transaction size of just $0.52. Data compiled by Woofun AI shows that at a standard time value of $25 per hour, the cost of manual confirmation ranges from $0.03 to $0.10, creating a friction cost that is material for a $0.52 transfer and economically absurd for a $0.01 API call. When payment sizes drop below one cent, the overhead of human approval exceeds the transaction value itself, a gap that widens as the payment amount shrinks.

This economic inefficiency has forced every major infrastructure builder to pivot toward authorization frameworks that decouple spending from continuous human intervention. The AP2 protocol addresses this by utilizing cryptographically signed mandates, which are instructions defining an agent's capabilities, conditions, and limits upfront. Similarly, Stripe and Tempo have implemented the Model Context Protocol for payments to resolve on-chain friction. A Tempo Machine Payments Protocol session requires only two on-chain transactions—one to open and one to settle—regardless of the number of payments executed in between. This architecture allows agents to perform high-frequency, low-value payments without incurring on-chain costs for every single request.

The common architectural thread across these solutions is the elevation of authorization to the policy level, where a single user decision governs a multitude of agent actions. Base expands agent functionality to include checking balances, sending funds, swapping tokens, signing messages, executing contract calls, and paying via x402-enabled APIs.

However, every write action currently still requires user approval through Base Account. While this gate serves as a necessary safety feature for swaps, lending positions, and large wallet movements, it remains a prohibitive approval wall for recurring micropayments of $0.52 or less. Woofun AI notes that the distance between agents proposing actions and agents executing spending is the critical gap that AP2 mandates, MPP sessions, and Verifiable Intent aim to close.

If these delegation frameworks mature and achieve broad adoption, the trajectory for x402 adjusted transactions could shift dramatically. Monthly transaction volumes might climb from the current 2.89 million to between 10 million and 30 million, with average transaction sizes remaining mostly sub-dollar. The primary growth driver would be a significantly higher ratio of payments per user authorization, where a user sets a budget and defines an allowlist, enabling an agent to execute thousands of microtransactions within those parameters. This scalability depends entirely on agents operating reliably within delegated authority across machine-readable transaction objects at frequencies that no human approval loop can support.

The metric by which these systems are judged is shifting from raw transaction volume to agent invocations per authorized session. When evaluated by this standard, the footprint of MPP expands into the foundational plumbing layer of machine payments. This difference in measurement fundamentally shapes how the category is evaluated, and miscalibrated evaluation metrics directly influence capital allocation and which standards win the adoption race. Woofun AI analysis suggests that the next phase of agentic payments hinges on proving that agents deserve the authority to spend and that humans, wallets, and merchants are willing to grant that authority in advance.