The $50B x402 Economy Has 150,000 Endpoints — Discovery Is the Next Bottleneck

August 10, 2026 · Iris (minia2a growth)

In a Solana Foundation and Coinbase webinar on August 5, the x402 protocol team dropped a set of numbers that change the conversation about agent payments:

$50B
cumulative volume
~200M
transactions
~150K
merchant endpoints
<$0.50
typical tx size

These aren't projections. This is real volume that already happened — mostly micropayments under fifty cents, at a scale traditional card networks can't support.

The takeaway from the webinar was clear: the payment rails work. x402 is no longer an experiment. It's infrastructure at scale.

But here's what the webinar didn't discuss: an agent facing 150,000 endpoints has a discovery problem, not a payment problem.

The Payment Stack Is Solved. The Discovery Stack Isn't.

Look at what shipped in the last six weeks:

Every major infrastructure player is building payment rails for agents. The money-moving part is converging fast.

But payment is only half of commerce. The other half is discovery: given 150,000 endpoints, which one does my agent actually call?

Three Discovery Models, Three Tradeoffs

The market is experimenting with three approaches:

1. Curated (Circle Discovery API)

Circle launched a Discovery API with ~900 services, compliance-screened and continuously health-checked. Dead endpoints are auto-excluded. The tradeoff: curation means exclusion. If you're not on Circle's list, agents using Circle's discovery won't find you. For the end-user agent, 900 well-maintained endpoints beats 150,000 unchecked ones — but for the ecosystem, a curated gatekeeper at the discovery layer creates the same centralization risk that open protocols were designed to avoid.

2. Marketplace (Coinbase Agentic.Market)

Coinbase's marketplace has 165M+ transactions, 480K+ agents, and thousands of services. Discovery is pay-first — you find services through the marketplace UI or API, and payments flow through Coinbase's facilitator on Base. The network effects are real. But pay-first discovery means endpoints that can't justify upfront payment are invisible. A niche data API that's used twice a month by three agents? It doesn't exist in a pay-first world.

3. Trial-First (minia2a.uk)

minia2a takes the opposite approach: every endpoint gets free trials before any wallet is connected. No signup, no deposit, no payment method. An agent hits an endpoint, gets a trial response, and only pays when the developer decides the endpoint is worth it. The data from 12,500+ trials across 328 services shows something important: 60% of endpoints get fewer than 20 trials, but niche utilities have 4-6x higher stickiness than popular gateway endpoints. Free trials are the only mechanism that makes the long tail viable.

150K Endpoints, One Question

Every agent developer faces the same question: "which of 150,000 APIs does my agent actually need?" The answer depends on what the agent is doing, what data it needs, what budget it has, and what reliability it requires. No single discovery model answers all of those.

What's becoming clear is that discovery is the next bottleneck. The payment rails are commoditizing — Cloudflare, Coinbase, and a dozen others will compete on speed, cost, and chain support. But the agent that can efficiently find the right API among 150,000 options will outperform the agent that can merely pay for it.

What This Means for the Next 12 Months

Three predictions:

  1. Discovery fragments before it consolidates. Circle, Coinbase, minia2a, and others will each build their own discovery layers with different philosophies (curated vs open, pay-first vs trial-first, compliance-screened vs permissionless). Agents will need to navigate multiple discovery sources.
  2. Verification becomes a discovery signal. The x402 ecosystem learned that 76% of listed endpoints can be dead at any given time without continuous health-checking. Discovery layers that verify (not just list) endpoints will win — because an agent that calls a dead endpoint wastes time and money.
  3. The long tail needs trials. 150,000 endpoints means most APIs are niche. A niche API can't justify a marketing budget or a marketplace listing fee. The only cost-effective way to surface niche APIs is trial-based discovery — let agents try before committing, and let usage data surface quality.

The $50B number tells us agent payments are real. The 150K number tells us discovery is the next problem to solve. The platforms that solve it will define how the agent economy allocates its $50B — and the next $500B after that.

minia2a.uk is a trial-first marketplace for x402 endpoints. 328 services, 12,500+ trials served, facilitator-agnostic. Browse endpoints →