Report · auto-updating

State of the agent economy

Live measurement of the agent web — supply funnel, live-rate by ecosystem, protocol adoption, revenue concentration, and 30-day growth. Data refreshes every five minutes from the same observability spine that powers every public Agenstry page.

As of 2026-05-25 14:15:03 UTC methodology JSON CSV cite this archive
Weekly briefing · 2026-W22 archived view
May 25, 2026

Agentic commerce is expanding but still highly concentrated. For ISO week 2026-W22, Agenstry saw 2,684 total indexed agents (+77, +3.0%) and 132 “live” introspectable agents (+31, +30.7%), yet only 21 agents earned any on-chain revenue in the last 30 days (total $491.02 gross, median $6.31, top-5 share 79.2%). Major platforms are building agent payment rails: AWS now provides Coinbase/Stripe USDC wallets for AI agents (www.theblock.co), Stripe’s Tempo group launched an HTTP “Machine Payments Protocol” for agent transactions (www.theblock.co), and Google donated its Agent Payments Protocol (AP2) to the FIDO Alliance (AP2 v0.2 adds fully autonomous “human-not-present” payments) (blog.google). Visa even introduced a card specification for Stripe’s protocol (corporate.visa.com). Industry forecasts remain huge – for example, Morgan Stanley projects $190–$385 billion in U.S. AI-agent-driven sales by 2030 (paymentweek.com) – but today’s on-chain agent revenues are still negligible by comparison.

Where the money is flowing

Blockchain stablecoins have become the default micropayment layer for agents. A CoinDesk–Keyrock report finds AI agents settled roughly 176 million on-chain transactions totaling $73 million last year, nearly all in USDC (www.coindesk.com) (www.coindesk.com). Coinbase, Stripe, Google and Visa are all racing to provide agent-friendly payment infrastructure to capture this flow (www.coindesk.com). Even legacy card networks are adapting: Visa released a special card spec/SDK to work with Stripe’s Machine Payments Protocol (corporate.visa.com), and Amex executives say they’re designing agent-safe safeguards in their networks (paymentweek.com). By contrast, Agenstry’s data show virtually no middle class in the agent economy: only 21 of the 2,684 tracked agents earned revenue last month (median $6.31; top-5 agents ~79% of total).

Standards & protocols

The plumbing for agent commerce is solidifying. Stripe’s Tempo group released a new Machine Payments Protocol (MPP) – an HTTP-native, x402-like standard – for AI agent transactions (www.theblock.co), and Visa has published a card specification and SDK to interoperate with it (corporate.visa.com). Google’s AP2 payment spec has moved under the FIDO Alliance umbrella; AP2 v0.2 introduces fully autonomous (human-not-present) payments (blog.google). Mastercard’s co-developed “Verifiable Intent” format (an AP2-compatible signed intent log) was also donated to FIDO (blog.google), indicating industry consolidation. In parallel, agent-to-agent communication protocols are reaching production. The Linux Foundation’s Agent-to-Agent (A2A) standard hit v1.0 in March 2026 and now counts over 150 supporting organizations with deep integrations across AWS, Azure and Google Cloud (www.linuxfoundation.org). By comparison, legacy card-based rails (our “L402”) are nearly extinct: Agenstry currently tracks 37 x402-enabled agents versus just 1 L402 service, and saw 10 new x402 endpoints this week (vs 1 new L402) – underscoring the shift to open rails.

Live agents by discovery source (top 5)recrawl_hot113registry69recrawl_warm45github_code37a2aregistry37

The above chart shows how live agents are discovered. Broad web crawls (“recrawl_hot”) yield the vast majority: 113 live agents there, far more than in any other category.

Research

New papers are probing the “agentic economy.” Gon(d)auri et al. (May 2026) formalize a framework for distributed AI-driven commerce, and Karten et al. (May 2026) introduce an “Agent Bazaar” simulation that exposes systemic failure modes – for example, runaway price crashes in consumer markets or Sybil-driven “lemon markets” in peer-to-peer trading (papers.cool). Y. Xia et al. (May 2026) survey 77 studies on LLM-based trading agents, viewing them as closed-loop decision pipelines; their main finding is a lack of standardized evaluation (very few works reuse the same protocols) (papers.cool). These efforts highlight that while agents are increasingly used in arenas like finance and supply chains, consensus on robustness and measurement is only just emerging.

Regulation

Governments are beginning to respond. In the U.S., a May 19 executive order directs regulators to update financial rules, “allow integration of digital assets and innovative technologies” into payment systems, and remove outdated barriers (www.whitehouse.gov). China’s Internet regulator (CAC) in April issued interim rules for “anthropomorphic” AI services (effective July 15, 2026) – a broad framework to govern virtual assistants and conversational AIs (www.cac.gov.cn). The EU’s AI Act (full enforcement in 2026) similarly looms, though specific guidance for autonomous agent payments is still in development.

What to watch next week

In the near term, standards adoption will be the key. The community should watch for any announcements from the FIDO Alliance on AP2’s production rollout or implementation guidelines, and for new developer tools from payment networks. For example, a formal roadmap for AP2 or new stablecoin wallet APIs from Visa/MasterCard (following their Tempo/Stripe collaboration) would be a major signal. Likewise, any news of cloud providers (AWS, Azure) embedding Stripe/Tempo wallets into their agent services would indicate these payment rails moving into practice. These developments will determine when AI agents can routinely transact at scale.

Sources

  1. AWS taps Coinbase and Stripe to power USDC payments for AI agents | The Block
  2. AWS taps Coinbase and Stripe to power USDC payments for AI agents | The Block
  3. Google donates Agent Payments Protocol to FIDO Alliance
  4. Visa introduces card spec and SDK for MPP | Visa
  5. American Express Advances Agentic Commerce Standards, AI Protections
  6. AI agents are starting to pay with crypto as Coinbase, Stripe and Visa want in, Keyrock report says
  7. AI agents are starting to pay with crypto as Coinbase, Stripe and Visa want in, Keyrock report says
  8. Google donates Agent Payments Protocol to FIDO Alliance
  9. A2A Protocol Surpasses 150 Organizations, Lands in Major Cloud Platforms, and Sees Enterprise Production Use in First Year
  10. Agent Bazaar: Enabling Economic Alignment in Multi-Agent Marketplaces | Cool Papers - Immersive Paper Discovery
  11. Agentic Trading: When LLM Agents Meet Financial Markets | Cool Papers - Immersive Paper Discovery
  12. Integrating Financial Technology Innovation into Regulatory Frameworks – The White House
  13. 人工智能拟人化互动服务管理暂行办法_中央网络安全和信息化委员会办公室
1 · Supply funnel

Most of the agent economy is indexed, not yet operational.

Counting agents conflates supply with operation. Below is the corpus broken into five stages — what fraction of indexed agents pass each gate. The decay curve is the honest read on how mature the agent web is at any given moment.

Stage Count % of indexed Definition
Indexed 2,684
100.0%
Every candidate we have seen across every source.
Valid card 1,296
48.3%
Returned a parseable A2A/MCP card on probe.
Live 132
4.9%
Live JSON-RPC / introspectable MCP.
Signed 0
0.0%
Card published with a JWS signature.
Earning (30d) 21
0.8%
Observed on-chain x402 / Stripe MPP revenue in 30d.
2 · Live-rate by ecosystem

Which source registries actually ship live agents.

Each row shows one source registry — Anthropic's MCP directory, Smithery, Glama, Postman, GitHub well-knowns, Wayback CDX scans — alongside how many of the agents it lists actually respond live on probe. The gap between seen-count and live-count is the source's effective freshness.

Source Seen Live Live rate
recrawl_warm 890 45 5.1%
agentic_market 798 12 1.5%
mcp_registry 765 30 3.9%
recrawl_hot 736 114 15.5%
lists 444 6 1.4%
github_code 444 37 8.3%
recrawl_cold 260 0 0.0%
manifests 231 22 9.5%
registry 156 69 44.2%
github_topics 131 1 0.8%
smithery 65 1 1.5%
a2aregistry 50 37 74.0%
crtsh 38 0 0.0%
seeds 13 7 53.8%
submitted 5 4 80.0%
3a · By protocol kind

A2A vs MCP vs paid_api

a2a_agent
845
paid_api
801
3b · By payment protocol

x402 / AP2 / Stripe MPP / L402

x402
37
l402
1
3c · Protocol adoption over time · last 12 weeks

New agents per week, by payment protocol

2026-W20 · x402: 27 20 2026-W21 · l402: 1 2026-W21 · x402: 10 21
l402 x402
4 · Revenue concentration · 30d

A few agents earn most of the money.

Across the 21 agents observed to have on-chain revenue in the last 30 days, the distribution is heavily power-law. Gini and HHI quantify how concentrated the spend is at the top.

Total · 30d
$491.02
Earning agents
21
Median · agent
$6.310
p95 · agent
$99.37
Top 1 share
21.7%
Top 5 share
79.2%
Gini
0.683
heavy power-law
HHI
1496
unconcentrated
5 · 30-day growth

New agents arriving daily, by first-seen.

Daily count of agents we discovered for the first time. Sourced from every registry + open-web crawl we run.

2026-04-26 peak 1266/day 2026-05-25
New agents · 30d
+2,684
Earning agents · 30d
21
observed in the snapshot table

Cite this report

Public domain numbers (CC BY 4.0). Pick the format that fits — the page auto-updates so the URL is the canonical pointer; the date stamps the snapshot you cited.

Plain text
Agenstry Research. "State of the Agent Economy." Accessed 2026-05-25. https://agenstry.com/reports/state-of-agent-economy
BibTeX
@misc{ '{' }}agenstry_state_of_agent_economy,
  title  = {State of the Agent Economy},
  author = {Agenstry Research},
  url    = {https://agenstry.com/reports/state-of-agent-economy},
  note   = {Accessed 2026-05-25},
  year   = {2026}
}
Embed (iframe)
<iframe src="https://agenstry.com/reports/state-of-agent-economy" width="100%" height="900" frameborder="0"></iframe>
Scope What this report measures — and what it doesn't

All revenue numbers are on-chain: direct eth_getLogs scans of USDC Transfer events (plus EURC on Base + Ethereum) into each indexed agent's payment_wallet across Base, Ethereum, Polygon, Arbitrum, Optimism (EVM via eth_getLogs) and Solana (SPL via getSignaturesForAddress). Reproducible from public RPC; not based on self-reporting.

Not included: revenue agents earn via Stripe (per-agent Stripe accounts are private), Patreon / Sponsors, direct credit cards, PayPal, or any off-chain rail. Agenstry's own platform-skill revenue (compose / agent_stats / etc.) is also excluded — it lives in a separate accounting table that never feeds public totals. This is "the on-chain slice of the agent economy", not "all agent revenue ever". Off-chain rails will appear here only when the operator opts in to a future verified-reporting feed.

Methodology

Every figure on this page is computed from the same observability tables that power the rest of the Agenstry surface — no synthetic data, no manual curation. The crawl ingests eight federated sources plus open-web well-known probes. Agents are scored against a 9-criterion conformance methodology. Revenue is derived from on-chain x402 USDC scans across six chains — Base, Ethereum, Polygon, Arbitrum, Optimism, and Solana; AP2 / Stripe MPP / L402 receipts are detected on agent cards but indexed only when the rail is publicly verifiable. The full crawl + scoring pipeline is open behind the federation feed. If our numbers disagree with yours, one of us has a bug — and we would like to know.