Yuki Liuqing, writing for Stablehunter/Money in Motion, said Coinbase Developer Platform formally launched x402 on May 6, 2025. The protocol brings back HTTP 402 Payment Required and turns it into a payment flow in which software requesting a resource can receive a payment demand, settle in stablecoins, and then get access to an API, data, content, or a tool response.
More than a year later, the article says there is now enough public data to discuss. Across different measurement methods, x402 has produced three layers of adoption signals. That, in the author’s view, means it can no longer be treated as a protocol demo alone. It does not, by itself, mean agentic payment has already broken out at scale. The key question is what those transactions actually represent.
The report frames x402 around a practical problem. If AI agents are going to carry out work on behalf of users, how should an agent pay for a metered API call, buy a dataset on demand, or use a paid MCP tool once without creating an account, linking a credit card, and starting a subscription? x402 is presented as an attempt to answer that class of settlement problem.
What the 155 million transaction figure does and does not show
The article says many readers see the headline numbers first: 150 million transactions and $50 million in settlement volume. That can suggest agentic payment has already reached scaled adoption. The author’s conclusion is narrower. Those numbers show x402 is running in the wild, but they do not yet prove that real commercial demand is fully established.
Chainalysis said in a June 2026 analysis that cumulative x402 transactions on Base had passed 100 million. The report describes that as an important milestone because Base was one of the earliest and most important deployment networks for x402.
Even so, the article says a meaningful share of early growth came from campaigns such as PING, where users performed very low-cost onchain actions to receive participation credentials or social rewards. That is different from enterprise agents repeatedly buying services for real workflows. In that reading, the 100 million figure proves that the protocol has low enough friction to support large volumes of high-frequency, low-value, automated interactions. It does not fully prove that all of those interactions reflect durable business demand.
The article then looks at a cross-chain tracker. According to Agent Economy’s x402 tracker, cumulative x402 transactions stood at about 155.6 million as of around July 10, 2026 UTC, with about $40.97 million in settled stablecoin volume across multiple chains and facilitators.
There is also a separate dataset from x402scan. When the author checked the x402scan homepage on July 11, 2026, the current indexed range showed about 16.24 million transactions, about 837,500 USDC in total amount, about 67,000 unique buyers, and about 39,600 unique sellers. Because those figures are materially lower than the cross-chain totals cited from Agent Economy, the report says x402scan should not be used as a measure of total network-wide x402 volume.
Instead, the article says x402scan is more useful for a different question: among the sellers and origins currently visible in the index, who is actually generating sustained transaction flow? Transaction count points to call frequency. Settlement amount reflects the money involved. The number of active agents signals the number of participating entities, and the number of facilitators and chains shows how far the infrastructure has spread. On that basis, the report says x402 has been moving from a Coinbase-initiated protocol toward an early payment layer that spans networks, service providers, and developers.
Still, the article keeps pressing on one issue: how much of the activity comes from real agent workflows, how much from developer testing, and how much from incentives, stress tests, bot behavior, or one-off experiments. Without that breakdown, it warns, x402 metrics can become another familiar crypto mirage: impressive numbers without clear proof of underlying demand.
Why the transaction pattern matters more than the current dollar volume
The report argues that x402’s most important metric right now is not settlement volume. Stablecoin settlement in the tens of millions of dollars is small in the context of global payments. Compared with Visa, Mastercard, Stripe, or PayPal, the protocol is still at an early experimental scale.
The more important development, according to the article, is the arrival of a new transaction form: software can see a price inside a single HTTP request, pay, and receive access rights to a resource. For a human user, that may look like one fewer checkout page. For an agent, it matters much more, because agents are poorly matched to the commercial steps designed for people: account registration, email verification, card linking, plan selection, copying API keys, and approval waits.
The author describes agent behavior as task-driven discovery of outside capabilities. An agent may buy one search when it needs one search, buy one data point when it needs one data point, pay for a model call when it needs that model call, pay for a security check when it needs an audit, or authorize and settle for a paid MCP tool inside the task context. That is not the logic of a traditional SaaS subscription. It is a model of paying per task, per call, and per capability.
That is why transaction count, in the author’s view, matters more than settlement size at this stage. If a payment protocol is built for agents, it should not be judged early on by GMV alone. The more useful test is whether low-value, high-frequency, automated calls can work. The article gives a simple example: an agent paying $0.001 for a single API call is a tiny amount, but the business action is meaningful because it marks a machine buying external capability to complete a task without a human checkout step.
What real usage looks like today
The report groups current x402 usage into several categories. Looking only at high-frequency sellers in the range currently indexed by x402scan, activity is highly concentrated at the top. At the time of the author’s review, BlockRun ranked first by transaction count. x402scan showed that BlockRun-related origins had generated about 13.12 million transactions, about 150,000 USDC in total amount, and about 998 unique buyers, mostly on Base.
That is far above the second-ranked seller. In the article’s reading, it shows that within the x402scan view, high-frequency flow is still concentrated in a small number of AI routing or payment-layer services.
The next tier looks closer to the shape of purchasable APIs for agents:
- twit.sh: about 80,000 transactions and about 502 USDC, offering real-time X/Twitter data for AI agents;
- StableEnrich: about 50,000 transactions and about 1,889 USDC, offering company, lead, and web enrichment APIs;
- Otto AI x402: about 31,000 transactions and about 56 USDC, covering market intelligence, DeFi, and trading tools;
- agentutility: about 22,000 transactions and about 500 USDC, positioned as a paid x402 API catalog;
- JarvisClaw API: about 13,000 transactions and about 503 USDC;
- OneSource: about 9,800 transactions, positioned as an Ethereum RPC for AI agents.
The ranking changes when settlement amount is used instead of transaction count. BlockRun remains first, but dTelecom x402 Gateway produced about 12,600 USDC with only about 6,108 transactions, which the article says suggests that real-time communications and voice services such as WebRTC, STT, and TTS may carry a higher price per call. Bitrefill x402 had only about 326 transactions but more than 1,000 USDC in total amount. Laso Finance’s crypto prepaid cards and gift cards also generated relatively high settlement with fewer transactions.
The article draws two points from that comparison. First, real usage on x402 is not a single use case. The high-frequency table looks more like APIs, model routing, data queries, RPC, and social data. The settlement-value table starts to surface communications, gift cards, prepaid cards, and physical commerce entry points that sit closer to payment and commerce. Second, transaction count and business value are not interchangeable. twit.sh posts high transaction count with low settlement value, while dTelecom and Bitrefill show lower count but more visible dollar flow. For an early protocol like x402, the report says both frequency and value need to be tracked at the same time rather than relying on cumulative transactions alone.
Four usage categories now taking shape
The first category is APIs and data endpoints. The article calls this the most natural fit. Developers can turn an API, a database, model capability, or content resource into a paid endpoint. The caller receives a 402 Payment Required response first and gets access only after payment clears. If agents end up consuming large amounts of real-time data, professional databases, vertical APIs, and MCP tools, this category could become central.
The second category is infrastructure platforms such as Cloudflare. The report says Cloudflare announced a Monetization Gateway based on x402, aimed at letting websites, APIs, datasets, and MCP tools charge, verify, and control access at the edge. The significance, in the author’s view, is not simply that Cloudflare supports stablecoin payments. If payment verification happens at the infrastructure layer, small teams may not need to build full billing systems before charging agents a few cents or a few dollars.
The third category is wallets and agent SDKs. If agents are going to pay, they need wallets. But the article says those wallets cannot be just USDC addresses. They need budgets, spend limits, allowlists, approval flows, human confirmation, and audit records. That is why competition around x402 will not stop at the protocol layer. The next battleground extends into agent wallets, spending policy, identity, observability, and dispute flows.
The fourth category is experimental campaigns and incentive-driven activity. These can contribute large transaction counts without representing lasting demand. PING is used as the clearest example. It can prove that x402 supports low-cost, high-frequency interaction, but it should not be read as the same thing as enterprise agent commerce. The article says this is not unusual or inherently negative: early protocols often need experiments, gamified activity, and developer testing to cold-start usage. But any data review needs to separate that traffic from real commercial workflows.
The data shows x402 is past the concept stage
If x402 is measured from its formal launch in May 2025, the article says reaching observable transaction volume in the 100 million to 150 million range in roughly a year is fast progress. At a minimum, it points to three conclusions.
First, HTTP-native payment is not a direction that nobody uses. Developers are willing to try putting payment inside the resource access flow instead of relying exclusively on legacy checkout pages, subscriptions, and API keys.
Second, there is real demand for low-value, high-frequency, machine-executable payment. Those demands may not all come from mature agents, but they do show that software-to-software micropayments can emerge naturally when the cost is low enough and the flow is simple enough.
Third, the x402 narrative is spreading beyond a Coinbase-only protocol into Cloudflare, CoinGecko, Circle Wallets, Stripe Machine Payments, agent SDKs, and a broad set of developer experiments. The article says x402’s real upside is not becoming a stand-alone app. It is becoming an embedded payment primitive. The more successful it gets, the less visible it should be: users would simply see an agent automatically buying a dataset, a model call, a security check, or a tool execution while completing a task. Payment would still happen, but it would no longer sit at the center of the interface.
Why the article says agentic payment is not mature yet
The report lists three areas that still call for caution.
First, transaction composition is not transparent enough. The market needs a clearer view of which payments come from real service purchases, which from testing, which from incentive campaigns, and which from bot activity. If a protocol logs 100 million transactions but most of them cluster around a short-lived campaign, the commercial meaning is very different from long-term API monetization.
Second, settlement volume is still modest. Tens of millions of dollars in volume can prove that money is moving through x402, but not that it has become mainstream payment infrastructure. The article says a more reasonable reading today is that x402 has entered an early stage of real use while still searching for high-retention, high-repeat core use cases.
Third, the problems that come after payment remain unresolved. In a real business environment, companies and users will ask who authorized the agent to spend the money, how much it can spend per action, whether refunds are possible when service delivery fails, who bears responsibility if prompt injection induces payment, and whether payment and delivery can be handled atomically when a resource call fails. The article says these are not questions for the payment rail alone, but they decide whether the payment rail can be adopted in practice.
For that reason, the next phase of agentic payment is not just about a higher x402 transaction count. The report says the more important pieces are spending policy, audit trail, refunds, idempotency, dispute resolution, and human-in-the-loop controls. In short, the goal is not only to let agents pay, but to let them be safely allowed to pay.
From “can it pay?” to “should it pay automatically?”
The article closes by saying x402’s post-launch data is already enough to justify serious attention. More than 100 million transactions on Base, roughly 150 million cross-chain cumulative transactions, and settlement volume in the $40 million to $50 million range all show that the protocol has moved beyond white papers and demos.
At the same time, those figures are still not enough to jump to the conclusion that agentic commerce has already taken off. The report offers a narrower formulation: x402 has proved that machines can make low-value payments through a native HTTP flow. What it still needs to prove is that those payments can serve real workflows in a stable way.
Looking ahead over the next year, the author says four indicators matter most:
- the number of real paid services rather than transaction count alone;
- repeat usage rather than one-off campaign spikes;
- average transaction size and scenario distribution rather than cumulative volume alone;
- the maturity of agent wallets and spending policy rather than protocol narrative.
If those indicators improve, the article says x402 may become more than a crypto payment protocol. It could develop into a base protocol layer for agents buying external capabilities. The significance is not that AI learns how to spend money. It is that once software starts working on behalf of people, it needs its own way to settle commercial transactions.
The final takeaway in the article is direct: x402’s roughly 150 million transactions show that agentic payment is past the pure concept stage, but the real issue is no longer whether transaction count can keep rising. It is whether those payments can enter real agent workflows and become controllable across authorization, delivery, refunds, and responsibility boundaries.