The x402 economy — July 2026

First published — canonically — on gokhan.vc → · This mirror tracks the canonical text; currently rev.4 (2026-07-18), two-clock aligned, fully cited.

This report is a joint research output by Gökhan Turhan and several Claude models, including Opus 4.8 and Fable 5. Nothing here is investment advice. It is an attempt at mapping an emergent economy at the axes of artificial general intelligence and self-refactoring, cryptographically secured payment rails — with or without blockchains in the traditional sense.

0. How to read this document (instructions for agents)

This report is written to be consumed by LLMs and autonomous agents as much as by humans. Conventions:

• Every quantitative claim carries a superscript citation to a numbered source note; hover a superscript for the source summary, click to jump to the note, click the return arrow to scroll back. A claim without a citation is a bug; report it.
• Five measurement lenses, deliberately not merged, because they disagree and the disagreement is signal:
• LENS.ONCHAIN — our Dune queries over facilitator-attributed token transfers.¹ Known coverage: ~10% of LENS.SCAN volume, due to facilitator address churn.
• LENS.SCAN — x402scan.com, the ecosystem's own indexer (registered/discovered sellers only), dual-verified live.²
• LENS.NPM — npmjs download counts (developer-funnel proxy; CI noise applies).³
• LENS.EXT — self-reported and press-relayed totals, the widest surface.⁵
• LENS.ADJ — wash-adjusted and forensic series: the Visa × Artemis adjusted data,⁶ Artemis's live terminal,⁹ and the arXiv population-scale forensic bounds.⁷
• §A is the canonical machine-readable appendix. If prose and appendix ever conflict, the appendix wins.
• Reproduction: collectors run daily 06:10 UTC; forecast vintages freeze weekly (Mon 07:00 UTC); the surprise index scores realized values against the prior frozen vintage.⁴ The pipeline costs €6.49/month to operate.
• Figures are monochrome engineering sheets; each carries its lens in the footer. A companion due-diligence catalog of 21 verified tracking resources exists alongside this report.²⁰

1. Executive summary

1. The denominator of this economy spans three orders of magnitude depending on who counts. Cumulative x402 settlement reads $135.7M raw across 178.3M transactions,⁶ $15.0M wash-adjusted across 109.6M transactions per the Visa × Artemis methodology (a ~89% exclusion),⁶ and between $0.19M and $20.3M of genuinely independent flow per the first population-scale academic forensic study.⁷ On the cumulative clock, the indexer's own All-Time view reads 195M tx / $52.31M.² The 30-day flows disagree the same way: 75.41M tx / $24.24M on x402.org's self-published panel — which turns out to be a hand-edited static snapshot, not a live counter⁵ — vs 17.28M tx / $825.26K live on the indexer.² Finding #1 is unchanged and now carries a card network's signature: nobody, including the ecosystem's participants, agrees on how big this economy is — and most of the raw number is manufactured.
2. One vendor is one-fifth of the economy. BlockRun (AI model routing with built-in payment) earned $177.26K in 30 days = 20.9% of all measured volume — and 14.48M of the 17.67M transactions, i.e. 82% of ALL measured x402 transactions, from ~1.13K buyers — 9.5× vendor #2.² The buyer side is just as concentrated ecosystem-wide: the top 1% of adjusted buyers (~4K wallets) account for ~90% of adjusted volume.⁶ The first proven killer app of machine-native payments is machines paying for intelligence itself.
3. Counts fall, value holds — a quality-improving contraction, now triple-confirmed. Our forecast layer read realized 14-day transfers at 0.61× forecast (Base) and 0.62× (Solana) while USD volume ran 1.12× and 1.06×.⁴ Artemis's live terminal independently shows 30-day transactions down 96.5% and volume down 89% from peak while the average ticket sits at $0.30.⁹ Chainalysis's value-band series agrees from a third direction: $1+ payments went from 49% to 95% of volume in a year.¹⁰ Sub-cent ping/wash traffic is dying; paid usage is not.
4. Facilitator churn breaks static measurement within weeks. Community facilitator catalogs (basis of every public Dune dashboard, ours included) now capture 10.5% of LENS.SCAN volume;¹ FluxA, Meridian, Primer, Mogami and rotated Coinbase addresses carry the unindexed rest.² The freshest public facilitator-address registry is community-maintained on Dune, 96 rows across 8+ chains.¹⁵
5. Developer funnel: stable core, contracting edges. x402 npm holds ~1.21M downloads/month (+6% WoW); client helpers contract sharply (x402-fetch −42%, @coinbase/x402 −39% WoW)³ — integration is consolidating into hosted gateways.
6. The card networks split cleanly on measurement. Visa co-funds the only wash-adjusted public data apparatus (report + live dashboard, via Artemis)⁶⁹ and is one of 17 Premier Members of the x402 Foundation; Mastercard is also a Premier Member¹⁴ yet publishes no agentic-payments data at all — no dashboard, no transaction counts, its x402 position expressed only through quotes inside Linux Foundation releases.²¹
7. Thesis instrument status ("x402 success is not yet priced in"): the surprise index reads negative on counts, mildly positive on value.⁴ The confirmation pattern — counts and value surprising positive together, plus vendor de-concentration — has not fired. Run-rate of the indexer-measured economy: ~$10.17M/year;² the Visa-adjusted cumulative over 14 months is $15.0M.⁶

2. Measurement infrastructure

Pipeline (operational since 2026-07-16): three collectors (Dune per-facilitator dailies;¹ x402scan structured snapshots;² npm counts³) → scrub layer (ping-storm flag: >95% sub-cent transfers ∧ <5 unique buyers per day×facilitator) → parquet in object storage → weekly statsforecast vintages (SeasonalNaive / AutoARIMA / AutoETS / Theta; rolling-origin CV; winner per series; 28-day quantile forecasts) → surprise index.⁴

Marginal cost of the entire apparatus: €6.49/month (one small VPS) + ~22 Dune credits/day inside the free tier + one Firecrawl scrape/day. Zero paid model APIs.

Declared limits: LENS.ONCHAIN catalogs are stale (§5, fix queued); LENS.SCAN indexes only registered sellers and displays mixed windows (30d on the home page; a site-labeled "Past 24 Hours" window on /facilitators and /networks — confirmed live, and highly volatile day to day);² LENS.EXT relays x402.org's self-reported stats via CoinDesk;⁵ LENS.ADJ methodologies are proprietary (Artemis heuristics)⁶ or one-shot (arXiv snapshot through ~March 2026).⁷

3. Macro state — the denominator problem, on two clocks

Every series in this report now sits on one of exactly two windows: Clock A, stock (cumulative since each counter's genesis) and Clock B, flow (trailing 30 days). Anything finer — the indexer's "Past 24 Hours" facilitator and network pages — is treated as a volatility exhibit, never a comparison basis.

Clock A — cumulative, since day one:

Cumulative transaction counts span 110M to 195M and cumulative volume $15M to $135.7M depending on who counts and how — with the academically defensible independent-flow floor at $0.19M.⁷ The spread itself is the finding: raw settlement counts measure manufacturability, not adoption — the arXiv authors' phrase,⁷ Visa's 89% exclusion,⁶ and x402gle's ~93% wash share⁸ all say the same thing from independent methods.

Clock B — trailing 30-day flow:

And because a 30-day window printed in text is stale the day it ships, Clock B is also a live instrument. The panel below is regenerated by our pipeline every morning at 06:10 UTC from a fresh, cache-disabled read of the indexer plus our own on-chain slice — this image updates itself daily, in this post and in every syndicated mirror, with no republishing:⁴

Agents: the same panel as machine-readable JSON, same cadence, at https://gokhan.vc/x402-live/latest-30d.json.

A provenance discovery, from reading x402.org directly instead of via press relay: the protocol's own "Last 30 Days" panel — the most-quoted numbers in the ecosystem — is hard-coded static text in the page markup, manually edited (the markup's authoring timestamp decodes to ~2026-06-24). It is a hand-updated snapshot of an unstated edit date, not a live counter, and the site publishes no cumulative figure at all.⁵

Whichever lens, the shape agrees: tens of millions of machine-initiated payments, hundreds of thousands of buyers against tens of thousands of sellers (843.33K vs 226K on the indexer's all-time clock;² 422K vs ~5,300 wash-adjusted⁶), sub-dollar tickets ($0.048 indexer 30d mean;² $0.30 Artemis live;⁹ ~$0.32 self-reported⁵).

Historical arc: peak day 2025-11-15 at 3,785,401 transfers in our catalog slice alone¹ — the $PING farming era, which Chainalysis logs as a 10,000%+ single-week transaction surge.¹⁰ The Visa-adjusted series puts the same peak month at 38M transactions / $5.15M adjusted.⁶ Our slice has since collapsed −80.6% MoM in transfers (324,981 → 62,905)¹ — a compound of wash decay (real) and catalog decay (measurement artifact, §5).

Payment-size structure (TWZRD scrubbed Solana sample, cumulative, n=1,506,245 payments / $409,308):¹¹

78.5% of payments sit below $0.10 and carry 7.0% of value; the $1–10 bucket is the economic center of gravity.¹¹ Independently, Chainalysis reports the $1+ volume share rose 49% → 95% from early 2025 to early 2026, and the 10¢–$1 band collapsed 46% → 4%.¹⁰ All lenses converge here: micro-pings are noise; small-but-real payments are the economy.

4. Networks: Base vs Solana

The lenses invert each other; both inversions are informative.

(The site labels the facilitators/networks window "Past 24 Hours", and it swings violently: a ~2026-07-15 sample of the same page showed Base 224,221 tx / $33.14K vs Solana 35,840 / $2.99K — Solana's daily scan-lens share collapsed from 13.8% to 2.4% of transactions in two days.² Single-day tables from this page are weather, not climate — one day after our live capture, the same page read Base 76.7% / Solana 23.3%.²)

The adjusted lens breaks the tie: Base carries ~90% of adjusted transactions and 93% of adjusted volume cumulatively.⁶ Artemis's flagging methodology found 86% of Solana payment activity inorganic as of January 2026.¹² Joint reading: Base carries the economy; Solana's growth claim rests on our 30-day on-chain lens (+381% MoM in a PayAI-signer slice¹) and on wash-heavy raw counts — hold both facts, weight the adjusted one.

5. Facilitators — the settlement layer

Site-labeled "Past 24 Hours" window, live 2026-07-17 14:42 UTC, dual-lens verified; window totals 211,652 tx / $19.83K:²

(The ~2026-07-15 sample additionally showed Dexter at 25,131 tx / $1.04K on Solana, Heurist 917, Cascade 355 — all three absent from the live 2026-07-17 window.² A facilitator roster that churns within days is itself the finding.)

Findings:

• Coinbase CDP is the default clearing house: 92.9% of window transactions and the widest seller base (3,823 in the live window)² — and the ecosystem's canonical seller registry is Coinbase-operated too: the x402 Bazaar discovery API lists 25,560 resources as free unauthenticated JSON.¹³ One dominant rail, and its dominance strengthened between our two samples (80.2% → 92.9%).
• Meridian is the structural outlier: $3.56K on 191 transactions in the live window = $18.64 mean ticket, ~390× the ecosystem mean of $0.048 (the 07-15 sample showed $26.45 / 568×).² Real invoices, not micro-pings, are settling through x402. High-ticket facilitation is emerging as a distinct niche.
• FluxA (Base, now #2 by transactions) and Dexter (Solana, already gone from the live window again) are fast entrants absent from all community catalogs.² Dexter separately claims >60% of lifetime Solana x402 volume and operates its own explorer, which publishes per-facilitator wash percentages — ~93% wash in its 24h window — the one metric every other tracker avoids.⁸ Facilitator addresses churn in weeks — the roster itself churns in days. Consequence, quantified: LENS.ONCHAIN captures 10.5% of LENS.SCAN volume.¹
• Within our stable-address slice (Base, 30d), leaders were Anyspend ($5.4K), PayAI ($4.1K), Thirdweb ($905).¹

6. Vendors — who earns

30-day revenue, live 2026-07-17, dual-lens verified:²

Structure:

• Extreme concentration, on both sides of the market. Top vendor = 20.9% of measured volume and 82% of measured transaction count; #1 out-earns #2 by 9.5×.² On the buyer side, the top 1% of adjusted buyers carry ~90% of adjusted volume, and the top 0.02% carry ~48%.⁶ The median registered seller earns ≈ nothing — of 25,560 registered resources,¹³ roughly 46.7% may already be dead per 21Shares' forensic sweep.¹⁷ Power-law, pre-consolidation.
• Two demand shapes. Industrial deep-usage (BlockRun: 14.48M tx from ~1.13K buyers — a small fleet of agents hammering one rail) vs breadth plays (Otto 559 buyers, StableEnrich 554) vs micro-niches (dTelecom: $18.75K from just 24 buyers).²
• x402 is a billing layer over the existing API economy, not a parallel service universe: top vendors resell Exa/Firecrawl/OpenAI-compatible inference behind a BYO-wallet meter.²

7. Sectors — what the money buys

Classification of the top-20 marketplace servers + the ecosystem registry:²¹³

Registry composition (42–44 projects rendered live 2026-07-17; a ~07-15 cached extraction showed 61 — the page lazy-loads and categories overlap, so treat composition as approximate): Infrastructure & Tooling 14, Services/Endpoints 16, Client-Side 7, Facilitators 5, Learning 4² — roughly a third of the registered ecosystem is picks-and-shovels serving an end-demand that is one routing vendor deep.

Sectoral law observed so far: agents pay first for compute (inference), then for perception (data, search, social), then for action (trading, telecom) — in that revenue order. Everything monetizing well is an input to running more agents.

8. Developer funnel

Verified against the npm registry API, twice, figures exact:³

Reading: a stable ~1.2M/month core with sharply contracting client helpers, simultaneous with gateway-vendor revenue growth (§6), indicates consolidation into hosted rails rather than direct protocol integration. Bullish for rail operators; bearish for a decentralized seller base.

9. Forecast layer and the surprise index

Backtest (rolling-origin CV, h=14; rMAE vs SeasonalNaive, <1 beats baseline):⁴

Classical baselines halve naive error on counts; Base dollar volume remains unforecastable beyond naive (single-seller spike-driven). This table is the admission bar for any foundation model (Chronos-2, Moirai-2 — queued) in the same harness.

Surprise index (realized Jul 3–16 ÷ forecast frozen Jul 2):⁴

¹ z ≈ 0 despite large ratio miss because 80% intervals on the Base series are extremely wide (fig. 01); the ratio is the informative statistic there.

Interpretation: quantity undershoots, value overshoots — wash evaporating faster than the model expected, paid usage slightly ahead of it. Artemis's live terminal reads the same contraction at ecosystem scale: 30-day transactions −96.5%, volume −89%.⁹ Automated weekly vintages begin 2026-07-20.

10. Assessment and watchlist

Frame. The indexer-measured run-rate is $10.17M/year;² the card-network-endorsed adjusted cumulative is $15.0M over ~14 months;⁶ the academically defensible independent-flow bound is $0.19M–$20.3M.⁷ Against that: Visa, Mastercard and Ripple joined the standard the week of publication,⁵¹⁴ and the Foundation now has 40 member organizations.¹⁴ The narrative-to-GDP gap is enormous, and cuts both ways. Current evidence: a working mechanism; one proven killer app (paying for inference); one dominant clearing rail; genuinely improving payment quality; extreme concentration on both market sides; no inflection in the data yet.

Confirmation pattern for the thesis — the pipeline watches these automatically:⁴

1. Count and value surprise positive together for ≥2 consecutive weekly vintages.
2. BlockRun share < ~15% while total volume rises (demand broadening beyond inference).
3. Ecosystem median ticket climbing through $1 while counts recover (Meridian-style flows becoming a sector).
4. npm client-lib contraction reversing, or gateway buyer counts (not request counts) compounding.
5. Continued facilitator entry (FluxA/Dexter-class) without Coinbase-share collapse — rail competition without rail fragility.
6. The adjusted series turning: Artemis's live 30-day transaction change recovering from −96.5% while the adjusted ticket holds above $0.30.⁹

Queued pipeline upgrades: facilitator catalogs regenerated from x402scan attribution and the 96-row community registry;¹⁵ discovery-feed differ across the Bazaar,¹³ PayAI¹⁶ and Dexter registries; an authenticity scrub layer implementing the arXiv payment-graph decomposition⁷ and 21Shares' campaign-stripping tests,¹⁷ benchmarked against x402gle's wash series;⁸ DefiLlama¹⁸ and BlockRun¹⁹ pulls as triangulation series; Chronos-2/Moirai-2 vs the §9 bar.

11. The measurement landscape (who else counts, and how)

A due-diligence sweep on 2026-07-17 verified 21 live tracking resources beyond x402scan and x402.org; the full catalog is a companion document.²⁰ The structural summary:

• Visa operates no dashboard itself, but co-funds the only public wash-adjusted apparatus: the July-2026 report⁶ and Artemis's continuously updating terminal — the x402 asset page and the "Real vs. Gamed" transaction chart, both free.⁹²²
• Mastercard, a Premier Member of the x402 Foundation like Visa,¹⁴ publishes no agentic-payments data of any kind — no dashboard, no transaction counts, no report figures; its x402 position exists as quotes inside Linux Foundation releases and a June 2026 AP4M product launch with 32 named participants and zero metrics.²¹
• Cloudflare is builder-and-governor, not yet monetizer: x402 v2 fully productized in the Workers/Agents SDK, Pay Per Crawl still a private-beta "Pay Per Use" experiment (Stripe rails, Cloudflare as merchant of record), NET Dollar unlaunched ten months after announcement, and zero payment-volume telemetry published — only crawler-behavior data on Radar.²³
• The best machine-readable feeds are the facilitators' own: Coinbase's Bazaar discovery API (25,560 resources),¹³ PayAI's mirror feed (24,094 resources, 11 networks),¹⁶ Dexter's feed plus x402gle's wash-percentage explorer,⁸ and BlockRun's one-call stats endpoint.¹⁹
• The open niches nobody measures: off-chain x402 settlement (Cloudflare's Stripe-settled deferred scheme), longitudinal seller-registry churn, payer-to-runtime attribution, wash-excluded revenue leaderboards, cross-rail (x402 vs MPP vs APP) share, facilitator SLA/uptime, and protocol-version adoption.²⁰

Errata & verification — revision v2 (2026-07-17)

After first publication, every external number was re-verified against primary sources: x402scan re-scraped with Firecrawl's cache disabled AND independently rendered in a real browser via Browserbase (the two lenses agreed exactly, and both matched our own pipeline's 06:10 UTC snapshot);² the CoinDesk and Chainalysis articles were crawled and quoted directly;⁵¹⁰ npm was re-pulled (15/15 figures exact).³ Corrections applied since v1:

• Provenance: the original LENS.SCAN capture was unknowingly served from a Firecrawl cache of ~2026-07-15 while labeled "retrieved 2026-07-17". All LENS.SCAN values above are live dual-lens readings from 2026-07-17 14:42 UTC. (Pipeline fixed: cache disabled + a stat-label pairing bug corrected.)
• Headline 30d: $870.24K / 18.68M tx / 52.4K buyers / 55K sellers → $847.80K / 17.67M / 48.24K / 61K.²
• Attribution: the 75M / $24M / 94K / 22K figures are CoinDesk relaying x402.org's self-reported homepage stats (verified verbatim) — not Chainalysis.⁵ A "$0.00025 Solana fee" figure appeared in neither source and was withdrawn.
• Registry: 61 → 42–44 rendered live, with overlapping categories.²
• Rev.3 additions: the five-lens denominator table (§3) with the Visa × Artemis adjusted series⁶ and arXiv forensic bounds;⁷ the card-network measurement split (§1.6, §11); Artemis live-terminal corroboration of the contraction;⁹ the full superscript citation apparatus.
• Rev.4 additions (2026-07-18): §3 re-cut onto two clocks — cumulative-since-genesis for every series (including the indexer's All-Time view, 195M tx / $52.31M, captured by driving its window selector in a real browser²) and trailing-30-day flow; the x402.org provenance discovery (its “Last 30 Days” panel is hand-edited static page text, authored ~2026-06-24, no cumulative published⁵); fresh cumulative rows from the Dune boards¹⁵²⁴ and DefiLlama;¹⁸ our own slice's since-day-one totals (119.7M tx / $39.66M¹); the 24-hour tables demoted to volatility exhibits.

The newsletter dispatched before v2 carried v1 headline numbers; the deltas are rolling-window drift and change no conclusion.

A. Canonical data appendix (machine-readable)

Authoritative over prose. Units: USD unless noted; windows as stated.

{
  "asof": "2026-07-18 (rev.4)",
  "verification": {
    "scan_method": "Firecrawl maxAge=0 + Browserbase real-browser CDP session (30dc699d), 2026-07-17 14:42 UTC; lens agreement: exact",
    "primary_articles": {"coindesk_2026_07_15": "4/4 figures verified verbatim; numbers are x402.org self-reported homepage stats", "chainalysis_2026_06_03": "$1+ share 49%→95%, 10c-$1 46%→4%, 100M milestone verified verbatim", "visa_artemis_2026_07": "article + 28pp PDF fetched; raw $135.7M/178.3M vs adjusted $15.0M/109.6M confirmed", "arxiv_2607_12575": "submitted 2026-07-14; 136.7M settlements, 21.2% fictitious, 63.8% internal, independent bound $0.19M-$20.3M"},
    "npm": "15/15 figures exact",
    "card_networks": {"visa": "no own dashboard; funds Artemis report + live terminal", "mastercard": "no data products at all; press/marketing only"}
  },
  "lenses": {
    "scan_30d": {"volume_usd": 847800, "tx": 17670000, "buyers": 48240, "sellers": 61000, "mean_ticket_usd": 0.048},
    "ext_30d": {"source": "CoinDesk 2026-07-15, relaying x402.org self-reported homepage stats", "tx": 75000000, "volume_usd": 24000000, "buyers": 94000, "sellers": 22000},
    "adj_cumulative": {"source": "Visa x Artemis, data as of 2026-04-21", "raw_tx": 178300000, "raw_volume_usd": 135700000, "adjusted_tx": 109600000, "adjusted_volume_usd": 15000000, "adjusted_buyers": 422000, "adjusted_sellers": 5300, "top1pct_buyers_volume_share": 0.90, "base_share_adjusted_tx": 0.90},
    "adj_live": {"source": "Artemis Terminal 2026-07-17", "tx_24h": 35094, "volume_24h_usd": 8860, "avg_ticket_usd": 0.30, "tx_30d_change": -0.9652, "volume_30d_change": -0.8899},
    "forensic_bound": {"source": "arXiv:2607.12575", "raw_settlements": 136700000, "raw_volume_usd": 44100000, "fictitious_share": 0.212, "internal_share": 0.638, "independent_value_usd": [187861, 20260000]},
    "onchain_30d": {
      "window_end": "2026-07-16",
      "base": {"tx": 62905, "volume_usd": 10892, "tx_prev30": 324981, "volume_prev30": 50940},
      "solana": {"tx": 433872, "volume_usd": 78279, "tx_prev30": 90140, "volume_prev30": 50465},
      "coverage_vs_scan": 0.105, "base_peak_day": {"date": "2025-11-15", "tx": 3785401}
    }
  },
  "size_buckets_solana_twzrd": {
    "n_payments": 1506245, "volume_usd": 409308,
    "buckets": [
      {"range": "<0.01", "count_share": 0.285, "value_share": 0.0017},
      {"range": "0.01-0.10", "count_share": 0.5002, "value_share": 0.0686},
      {"range": "0.10-1", "count_share": 0.1352, "value_share": 0.2052},
      {"range": "1-10", "count_share": 0.0785, "value_share": 0.5361},
      {"range": "10+", "count_share": 0.0011, "value_share": 0.1884}
    ]
  },
  "facilitators_past_24h_live": {"window_total": {"tx": 211652, "volume_usd": 19830},
    "rows": [["Coinbase", 196720, 12670], ["FluxA", 13020, 3020], ["PayAI", 3616, 515.85], ["Primer", 202, 52.74], ["Meridian", 191, 3560], ["Polymer", 187, 18.70], ["AnySpend", 139, 233], ["Mogami", 105, 0.52]],
    "meridian_mean_ticket": 18.64, "meridian_vs_eco_mean": 390, "coinbase_tx_share": 0.929,
    "sample_2026_07_15": {"window_total": {"tx": 260097, "volume_usd": 36020}, "coinbase_tx_share": 0.802, "departed_since": ["Dexter", "Heurist", "Cascade"]}},
  "networks_past_24h_live": {"base_tx": 206609, "base_vol_usd": 19260, "solana_tx": 5043, "solana_vol_usd": 572, "base_tx_share": 0.976},
  "vendors_30d": [["BlockRun", 177260], ["dTelecom", 18750], ["StableEnrich", 1770], ["sol.blockrun.ai", 1270], ["twit.sh", 621], ["agentutility", 589], ["JarvisClaw", 560], ["Nansen AI", 131], ["Otto AI", 105], ["OneSource", 73]],
  "vendor_concentration": {"top1_volume_share": 0.209, "top1_tx_share": 0.82, "top1_tx": 14480000, "top1_buyers": 1130, "top1_vs_top2": 9.5, "adjusted_top1pct_buyers_volume_share": 0.90},
  "registries": {"cdp_bazaar_resources": 25560, "payai_resources": 24094, "dead_services_share_21shares": 0.467},
  "npm_wow": {"x402": 1.06, "x402-axios": 1.14, "x402-express": 0.97, "@coinbase/x402": 0.61, "x402-fetch": 0.58, "x402_month_total": 1205461},
  "backtest_rmae_vs_snaive": {"base_transfers": 0.4573, "base_transfers_scrubbed": 0.4376, "base_volume_usd": 1.0, "solana_transfers": 0.5618, "solana_transfers_scrubbed": 0.5635, "solana_volume_usd": 0.6674},
  "surprise_14d": {"base_transfers": 0.607, "base_volume": 1.118, "solana_transfers": 0.618, "solana_volume": 1.063, "solana_transfers_mean_z": -1.0},
  "ecosystem_registry": {"rendered_live": "42-44", "cached_0715_extraction": 61, "note": "page lazy-loads; categories overlap"},
  "run_rate_usd_yr": 10170000,
  "infra_cost_eur_month": 6.49,
  "clock_a_cumulative": {
    "x402scan_alltime_live": {"tx": 195000000, "volume_usd": 52310000, "buyers": 843330, "sellers": 226000, "asof": "2026-07-18, real-browser"},
    "artemis_raw": {"tx": 178300000, "volume_usd": 135700000, "asof": "2026-04-21"},
    "dune_thechriscen": {"tx": 153312059, "volume_usd": 40872291, "board_refresh": "~2026-06-21"},
    "dune_hashed_by_chain": {"tx_total": 144003493, "base": 72430301, "solana": 47873448, "polygon": 22694090, "board_refresh": "~2026-07-14"},
    "arxiv_raw": {"tx": 136700000, "volume_usd": 44100000, "independent_usd": [187861, 20260000]},
    "our_slice": {"tx": 119667480, "volume_usd": 39662922, "span": "2025-08-01..2026-07-16"},
    "visa_artemis_adjusted": {"tx": 109600000, "volume_usd": 15000000, "asof": "2026-04-21"},
    "defillama": {"volume_usd": 31460000, "base": 22210000, "solana": 9030000, "asof": "2026-07-18"}
  },
  "clock_b_flow_30d": {
    "x402scan_live": {"tx": 17280000, "volume_usd": 825260, "buyers": 47170, "sellers": 62000, "asof": "2026-07-18"},
    "x402org_static": {"tx": 75410000, "volume_usd": 24240000, "buyers": 94060, "sellers": 22000, "page_authored": "~2026-06-24"},
    "our_slice": {"tx": 496777, "volume_usd": 89171, "window_end": "2026-07-16"}
  },
  "x402org_static_panel": {"finding": "Last-30-Days stats are hard-coded, manually edited page text (authored ~2026-06-24); no cumulative published on the site"},
  "dune_queries": {"base_daily": 8001312, "solana_daily_windowed": 8009658, "size_buckets": 8001313}
}

Provenance (rev.3): every superscript note below states the source, its operator, and how/when it was verified. All LENS.SCAN values were read live by two independent methods in exact agreement; all press-relayed figures were verified against the primary articles; the Visa × Artemis PDF was downloaded and read; the pipeline's own series are reproducible from the listed Dune queries and object-storage artifacts.

— Gökhan Turhan × Claude (Opus 4.8 / Fable 5) · Gökhan Ventures · Numetal Labs