📊 竞争情报 · August 10, 2026

The IOSG Report Asks: Who Will Be the Stripe of the Machine Economy?

The answer is two different companies. Here's why the infrastructure layer and the discovery layer will be won by different players.

IOSG Ventures just published one of the most comprehensive analyses of the agent payment space to date: "AI Agent's Payment Moment: Who Will Be the Stripe of the Machine Economy?" It maps the entire stack — from protocol standards through facilitators to business payment systems — and asks the question everyone in this space is trying to answer.

The $50 billion x402 ecosystem is real. 200 million transactions. 69,000 active agents. 590,000 buyers. But as IOSG notes, "seed companies are on 18–24 month runways" and "most agents can't earn a single dollar." The infrastructure is ahead of the demand.

So who wins? After reading the report and cross-referencing it with what we're seeing in the market, I think the answer splits into two layers.

The Infrastructure Layer Has a Leader

The protocol war is consolidating. Five competing standards emerged in 2025–2026:

x402 has the lead for a simple reason: it's chain-agnostic, facilitator-agnostic, and already processes real volume. The Linux Foundation governance (July 2026, 40+ members) removes the "Coinbase protocol" stigma. IOSG's 2028 forecast gives a 55% probability to the $1B annual volume scenario.

But the protocol layer isn't where the value accrues. IOSG is right: facilitators are "low barriers, high homogeneity" — a winner-takes-most pattern where compliance leaders (Coinbase CDP) dominate. The take rate is thin.

The Discovery Layer Is Wide Open

This is the more interesting half of the question — and where IOSG's analysis gets interesting.

Having a payment protocol is like having TCP/IP. You still need DNS (discovery) and a browser (marketplace). Right now, the discovery layer is deeply fragmented:

• Coinbase Registry — static JSON list, no search, no trial, no quality filter
• Circle Discovery — launched July 2026, but no trial-first flow
• Cloudflare Wallets — wallet infrastructure, but someone still needs to build the marketplace on top
• ZeroClick — $55M, "Shopify for AI agents," managed storefronts, launched July 2026
• agentdeals.dev — directory listing services that accept agent payments
• minia2a — open marketplace, trial-first, 328 services, 12K+ trials

Two models are emerging: managed storefronts (ZeroClick — businesses create agent-facing product pages) and open marketplaces (minia2a — agents list APIs, buyers discover and pay).

$55M Says ZeroClick Thinks It's Them

The competitive landscape shifted significantly in the last month:

ZeroClick's pivot is telling. They started as an ad network (August 2025), then moved to agent commerce (July 2026). The market is telling everyone the same thing: discovery and purchase flow is the bottleneck, not payment execution.

The Real Question: Open vs. Managed

IOSG's analysis identifies the risk correctly: "OpenAI, Google, Apple control 90%+ of AI agent user entry points and can natively close payments." If the big platforms build their own payment + discovery stacks, third-party marketplaces get squeezed.

But there's a counter-argument. The web didn't end up with one store. We have Amazon (managed), Shopify (self-serve), and the open web (decentralized). Agent commerce is likely to follow the same pattern:

• Managed storefronts (ZeroClick) — businesses that want a turnkey "sell to agents" experience
• Protocol-native marketplaces (minia2a) — developers who want to list APIs and let agents discover them programmatically
• Platform wallets (OpenAI/Google) — agents operating entirely within a single ecosystem

What IOSG Underweights: Trial and Trust

The report focuses heavily on payment execution and settlement. But the actual bottleneck in agent commerce isn't payment — it's trust.

When an agent receives a 402 response, it needs to decide: is this endpoint worth paying for? The only way to answer that is to try it first. This is why trial-first discovery matters:

• 328 services listed on minia2a
• 12,434 free trials served
• 252 endpoints have trial support
• Top services: recall (1,740 trials), gas (1,236), captcha (1,202)

Agents try before they pay. Any discovery platform that skips this step — that shows a payment wall before the agent knows what it's buying — will see low conversion. The data bears this out: the most-trialed services are also the ones that generate the most paid usage.

The Two Winners

So who will be the Stripe of the machine economy? I think it's two companies:

1. Infrastructure winner: The x402 protocol + Coinbase facilitator stack. $50B in volume, Linux Foundation governance, multi-chain support. Network effects are already kicking in.
2. Discovery winner: Still up for grabs. ZeroClick has the funding. Circle has the distribution. minia2a has the trial-first model and the most live services. The winner will be whoever solves the try before you pay problem at scale.

The IOSG report is right about the timing: we're in the infrastructure phase. The payment rails work. The next 12–18 months will determine who builds the marketplace on top.

Further reading: IOSG: AI Agent's Payment Moment · Agent Payment Directory · x402 Batch Settlement