Agent Payments This Year: Under the Halo, the Market Has Yet to Arrive

Original title: a year inside agentic payments: the uncomfortable truth

Original author: @13yearoldvc

Translation: Peggy

Editor's note: This article provides a relatively sober builder's perspective: over the past year, agentic payments have become a hot narrative at the intersection of AI, payments, and crypto, with companies like Stripe, Visa, Coinbase, and Google all making moves. Concepts such as stablecoin micropayments, x402, machine-to-machine settlements, and agent-powered e-commerce have also been heating up. However, after actually building products and engaging with merchants and developers, the author found that genuine demand has yet to materialize on a large scale.

The article breaks down several typical scenarios: agent-powered shopping is not inherently better than traditional e-commerce for most product categories because users still need images, comparisons, and browsing. Machine API payments seem suitable for stablecoin micropayments, but most developers currently solve this through subscriptions, prepaid credits, and existing billing systems. Agent-to-agent payments, while a long-term vision, remain in the early stages with a lack of real transaction volume.

Relatively speaking, agent-powered finance is one of the few areas where demand already exists. Funds, treasury teams, and DeFi users already pay for financial tools, and AI can deliver real capability improvements like real-time monitoring and automated rebalancing. However, this market also tends to favor established institutions that already possess licenses, compliance frameworks, and client relationships.

The author's final assessment is that what the agent economy truly lacks is not merely a payment layer, but more complex coordination capabilities—how to enable agents to collaborate with humans, verify task completion, and settle outcomes. Payment is just one part of that. For giants, early positioning is a defensive move. But for startups, what truly matters is finding markets that exist right now.

The following is the original text:

Over the past year, I have been building infrastructure for the agent economy and have been in conversations with teams at Stripe, Visa, Coinbase, Google, and dozens of startups advancing agent commerce. I mapped out the landscape, launched products, and tried to find the real market.

But the reality is: genuine demand hasn't arrived yet. For startups looking to enter this space, there are still many structural problems.

At its Sessions conference last month, Stripe launched 288 new products, and documentation related to agents now accounts for nearly 40% of all documentation page views. Its agent commerce marketplace has already onboarded over 1,000 merchants. Yet, at the Sessions event itself, the number of agents that actually registered and completed transactions was in the single digits.

Visa mentioned that its agent token currently requires 3 to 9 months of KYC approval, and it essentially requires a company to have at least $250 million in annual revenue to be eligible. Only companies of Amazon's or Walmart's caliber are capable of closing the identity verification loop today.

Coinbase reported that as of April, there were 69,000 active agents and 165 million transactions on x402. However, independent on-chain analysis shows real daily transaction volume of about $17,000, roughly half of which were test transactions (CoinDesk, March 2026).

What we learned building shop.fast.xyz

Agent to merchant, i.e., agentic commerce

We built shop.fast.xyz with the explicit goal of validating agentic commerce. Real goods, real merchants, real transactions.

But for most product categories, the current AI shopping experience is noticeably worse than traditional e-commerce. When buying clothes, electronics, or furniture, users want to see images, browse options, and compare side-by-side. Chatbot-style dialogue is actually a regression: you replace a rich visual interface with a text conversation. Humans shop with their eyes first.

Agents performed well on what we initially thought would be the hardest part. They understand what the user wants and handle requests like "something similar, but cheaper" pretty well. The model layer works. But it can't replace the experience of "looking at ten items at once and picking one." Chat interfaces can include product carousels and interactive displays, but at that point, you're essentially rebuilding an e-commerce front-end inside a chat window. For shopping scenarios requiring visual comparison, we haven't found a compelling answer for why the chat shell is better than the original e-commerce interface.

We did see demand on the merchant side, but it's more defensive in nature. Merchants want their stores to be discoverable by agents, not because many consumers are already shopping through agents today, but because they fear being left behind if agents become a mainstream channel. This presents the so-called Agentic Engine Optimization opportunity, but for now, it's a "nice-to-have," not a "must-have." Merchants are preparing in advance for a wave that hasn't yet arrived.

Where conversational commerce truly improves the experience is in high-frequency, low-decision-cost purchases where the user already knows what they want. The clearest example is ordering food. The market is large enough, the frequency high enough, and the decision fast enough, like "order pad thai from that place I liked last time." In this scenario, conversational agents might win. But the major food delivery platforms don't open their APIs. The only path is computer use, meaning having the AI visually operate the app like a human. This process is slow, fragile, and the inference cost doesn't justify a $15 lunch.

Another opportunity lies in online stores that are so complex they genuinely frustrate users—e.g., layered discounts, promo codes, membership points, chaotic checkout processes. An agent that can handle "help me apply coupons, use points, find the cheapest shipping, and complete the purchase in my language" could indeed simplify a currently broken shopping experience. This is particularly valuable for elderly users, non-native speakers, especially when cross-border shopping, or in very specific scenarios with highly niche and complex user needs.

But both of these opportunities require massive B2C distribution capabilities. You are competing with DoorDash and Amazon for user entry points. Distribution at a consumer scale is the strength of existing giants. The supply side of agentic commerce is ready, but the demand side is constrained by user experience and distribution channels, and more infrastructure won't solve those two issues.

What we learned from x402 and MPP

Agent to Web/API, i.e., machine commerce

We spoke with dozens of developers about their real payment needs. The pattern was almost completely consistent: current agent API usage is essentially recurring consumption, like compute, reasoning, and data sources. Developers already have subscriptions, API keys, linked accounts, and billing relationships with core service providers.

The typical argument for stablecoin payments is that the effective minimum cost for credit card payments on Stripe is about 2.9% plus 30 cents, making API calls under $1 uneconomical. But at today's low transaction volumes, prepaid credits solve the problem. Developers pre-fund their accounts, and the issue disappears.

The deeper problem lies in the supplier market. Most large SaaS companies don't want to offer fractional API access at sub-cent prices. Their business model is multi-year enterprise contracts. Companies that rely on large committed revenue streams will resist new pricing models that circumvent this.

Machine commerce is structurally a long-tail market. It serves small services, vertical data sources, independent developers, MCP servers, etc. Protocols like MPP and x402 are well-suited for this niche. But by definition, this is a market for professional users, and developers have historically been one of the most reluctant demographics to pay for services.

When Stripe Projects launched, it onboarded 32 service partners, including Vercel, Supabase, Cloudflare, Twilio, covering most of the core services developers use to build and deploy software, all accessible through existing billing systems. The top of the developer stack is already well-served. The opportunity for new payment rails lies beyond those top 30 service providers: it's real, but its scale is naturally smaller than the grand narratives suggest.

Content access follows the same logic. Agents are constantly scraping and summarizing articles, and publishers are starting to fight back. But when content monetization really scales, it will likely happen through CDN providers already sitting between publishers and the internet, like Cloudflare, which has launched AI audit tools, or through bulk licensing agreements between publishers and AI labs. The infrastructure opportunity will flow to existing players who already have distribution.

What we learned from agent-to-agent payments

Commerce between agents is the long-term vision, but remains almost entirely theoretical. No one has yet generated meaningful transaction volume. The truly difficult parts—agent discovery, trust establishment, term negotiation, and dispute resolution—are being tackled by various startups.

Once this transactional structure truly takes shape, it will look completely different from existing payment rails. Neither party has a human identity; latency requirements are sub-second; transaction amounts can range from fractions of a cent to millions of dollars; and it will involve multi-party settlements, unlike the bilateral buyer-seller model default in current payment rails. When it does happen, we believe it will explode at incredible speed and scale.

This is the core long-term bet for dedicated settlement infrastructure, and the bet is real. But a "real long-term bet" is not the same as the "current market." We were among those who spent months declaring this market would arrive, building a full infrastructure stack around it over the past few years, including our distributed network. Theoretically, it can scale to over 1 billion TPS with sub-50 millisecond latency and an average consensus time of 10 milliseconds. But we have to come back to where the market is today.

What we learned from agentic finance

Arguably, this is the only category with existing genuine demand. The customers exist and are already paying. Fund managers, treasury teams, and DeFi users already spend money on financial tools. Integrating AI into existing workflows is a natural product path.

Agentic finance will also create entirely new behavior patterns. An agent capable of autonomously monitoring and rebalancing hundreds of positions in real-time operates in ways humans cannot manually replicate. There are real capability improvements here, not just automation.

The challenge lies in the competitive landscape. Finance is a highly regulated industry that relies on existing relationships. Incumbents have licenses, compliance infrastructure, and client relationships. Startups can enter via less regulated areas like DeFi, or look for areas where incumbents are slow to move, or where AI creates new capabilities the giants don't yet possess. Overall, however, the competitive dynamics in this category favor incumbents more than the previous three, because adding AI to an existing product and customer base is much easier than starting from AI and building out the product and customers.

An honest summary

So, why is everyone still doing this? Two reasons.

First, incentives. Large companies have enough cash flow to bet on a future that may take years to materialize. For them, entering five years early is a rounding error; entering one year late could be catastrophic. So they have to do it.

Second, cognitive bias. When your business is payments, every problem looks like a payments problem. The agent economy needs a payment layer, so everyone builds a payment layer.

But payment is only part of a larger issue. The truly hard problem isn't moving money between agents, but how to coordinate work between agents and humans, how to verify that things are done, and how to settle outcomes. Payment is just a part of settlement. Settlement is just a part of coordination. And coordination is the real prize.

Large-scale coordination naturally generates a need for settlement mechanisms. Payments will become one instrument in this concerto, not the composition itself. The companies that truly solve coordination will eventually incorporate payments, rather than payment companies swallowing coordination.

Most existing giants are defensively building for a future of "machines trading at scale." For them, the timeline doesn't matter because they have near-infinite runways.

But startups don't have that luxury. We have to find where the market truly is right now. We can't just wait for the wave to arrive.

A year of building has led us in an unexpected direction. There is indeed activity there, growing fast, underserved. It exists outside the four categories we mapped out.

Original link