Maybe we all misunderstood the original intention of x402

Author: Charlie Liu

If you follow the fintech, crypto, or AI circles, you've undoubtedly seen the term "x402" pop up all over your screen.

A host of internet giants have jumped into the agentic payments space: Coinbase and Cloudflare have teamed up to launch the x402 Open Foundation, Google has incorporated cryptographic extensions into its AP2 standard, and even the usually reserved Adyen has rolled up its sleeves and entered the fray. Visa's TAP promises interoperability, and Stripe and OpenAI have teamed up to make ACP and "instant checkout" industry standards. A nearly forgotten internet status code—402 "Payment Required"—has unexpectedly become crucial in determining the business model of the next generation of AI agents. In fact, this undercurrent has been bubbling within the payments industry for over six months. When I wrote "AI Agents: The Next Frontier Reshaping the Future of Payments" on Substack in February of this year, I noticed that high-profile multi-million dollar seed rounds like Skyfire and Payman—backed by Web2 and Web3 payment giants like Visa and Coinbase—were already investing in startups in this field. Furthermore, Stripe, Visa, and Paypal all mentioned the promise of integrating AI agents with payments in their annual reports, albeit with limited space. But in recent weeks, x402 has truly gained traction, thanks to major AI and fintech players. Both x402 itself and the broader trend toward intelligent agent business have finally captured the attention of the broader tech and investment communities. Many commentaries have noted its origins in the HTTP 402 status code and analyzed how the rise of intelligent agent AI and the increasingly mainstream nature of crypto have fueled this discussion. However, they generally overlook a fundamental driving force: what is the true motivation behind this AI + crypto craze? It all started with the existential crisis facing AI companies. On the surface, everyone is talking about how "intelligent agents can finally make payments themselves," and imagining what new capabilities AI agents will unlock in the future. However, according to a closed-door conversation I had with a prominent Silicon Valley investor, the source of all this—the real pain point—is the survival pressure of large model manufacturers like OpenAI and Anthropic. While they pitch trillion-dollar futures to the capital market, they face a mountain of copyright lawsuits and growing public doubts about the ethics of their training data. If they can't establish a scalable, automated method to compensate for the data and databases they scrape, subsequent financing and massive capital expenditures will be under pressure. One theory I heard is that OpenAI's Sam Altman approached Coinbase's Brian Armstrong—both of them graduated from the YC academy. This may also explain why Coinbase briefly launched AgentKit earlier this year. It was a toolkit for developers that included support for the OpenAI SDK, giving intelligent agents trading capabilities. However, AgentKit didn't last long—it was quickly shut down. It wasn't due to a wrong intuition, but rather a divergence in their paths. The two companies' strategic focuses quickly diverged. Perhaps under pressure from its partnership with Perplexity and Shopify, OpenAI's growth strategy began to shift toward consumer e-commerce—becoming the interface through which people discover and purchase products—and therefore favored a checkout solution that merchants could adopt without major effort. Meanwhile, Coinbase, driven by the GENIUS Act and a series of US crypto regulatory actions, gradually shifted its ambitions toward internet money infrastructure. This naturally pushed it toward "machine-to-machine" traffic, where bots pay directly to pages or interfaces without creating accounts. Through first principles, we can see the four major camps formed by the giants. Although AgentKit was short-lived, it laid the foundation for Coinbase's subsequent upgrade, x402, and opened up new avenues for other major players, including [the company/enterprise]. It's precisely because of the aforementioned driving forces that, through this first principle, we can better understand the competitive and cooperative relationships formed by the various camps of the giants: Stripe × OpenAI: Dominating the consumer landscape and achieving a "smooth upgrade" OpenAI and Stripe jointly developed the ACP protocol and launched instant checkout in ChatGPT. This is essentially an "elegant upgrade" to the existing payment network. Stripe issues a shared payment token specific to a specific order and merchant. Agents use this token to complete transactions, while merchants continue to use familiar fraud prevention, refund, and tax processes—without reorganizing teams or retraining. Etsy has already implemented this, and Shopify is waiting in the wings. The core of this strategy lies in leveraging existing channels and usage habits: seamlessly integrating agent transactions into merchants' established payment pipelines. Coinbase × Cloudflare: Targeting the Machine Economy, Undertaking a "Restructure" Coinbase's upgraded version, x402, is even more ambitious. It makes the HTTP protocol's "402 Payment Required" status code executable: the server declares the price and accepted tokens, the client completes the payment (preferably USDC, the money-making machine on the Base chain), and proceeds with the transaction based on proof of payment. No account, login, or monthly billing required. This solution excels in micropayments between $0.01 and $0.10—for example, per-call API calls, per-article article context, and per-use analytics tools—resembling the initial demo of Skyfire, a company it invested in. Cloudflare's previously launched "pay-per-crawl" model paved the way for this, and Coinbase's strong entry has further united key players like AWS, Circle/USDC, and NEAR. Anthropic, in particular, is a perfect stand-in for the instigator, OpenAI. Unlike OpenAI, which focuses on consumer shopping, Anthropic's core strategy, the MCP framework, focuses on enabling intelligent agents to discover and invoke tools on a per-use basis. Its economic model naturally aligns with x402's "pay-per-request" model, far surpassing store-based payment methods. Google × Adyen: Setting Rules to Ensure Compliance and Auditability Google's AP2 standard aims to address the thorniest issue in the payment system: authorization. It uses a set of signed authorization instructions to bind user intent (such as price caps, frequency limits, and product category scope) to specific actions. The design is inherently payment channel agnostic and includes a cryptographic extension that allows agents to seamlessly connect to x402 links when processing very small payments, while maintaining a complete audit trail.

Furthermore, the participation of payment giant Adyen means that the system has enterprise-grade dispute resolution and compliance capabilities from day one.

Visa TAP: No New Rail, Inclusive

Visa's TAP does not attempt to rebuild payment rails, but rather plays a key compatibility role: helping issuers and acquirers identify agent transaction traffic and apply unified risk and dispute rules to it. The key lies in its interoperability positioning—TAP has explicitly stated its alignment with the Stripe + OpenAI camp's ACP and its interoperability with the Coinbase + Cloudflare camp's x402, effectively implementing its Network of Networks positioning and positioning itself as an invincible force. Add to that Mastercard's "Agent Pay," and you'll see that bank card brands are driving standard convergence rather than erecting walls around intelligent commerce. While these solutions may appear to operate independently, in practice they are forming a structure of division of labor, collaboration, and layered complementarity: Authorization Layer: The Google + Adyen camp's AP2 standard provides verifiable proof of user intent, serving as a "pass" for all operations.

• For consumer purchases, we use OpenAI and Stripe's ACP and traditional card networks for a smooth experience.

• For machine-to-machine micropayments, we use Coinbase and Cloudflare's x402 and cryptocurrency for minimal costs.