Blog
hedwigaihedwigaihedwigai

Payments

Who earns on Stripe and who earns on x402: verified startup revenue against agent payments

Neither rail publishes its merchants. Two sites publish what they can see: TrustMRR shows revenue that startups let it read straight from Stripe, RevenueCat or Paddle, and x402scan indexes every x402 payment, the stablecoin payments an agent makes per request, as it settles on Base or Solana. We read both on 12 November 2026, and the program that read them is one any MCP client can call, so every figure below can be fetched again.

In TrustMRR’s top 100 by monthly recurring revenue, 81 startups are paid through Stripe, for $10.13M a month. All of x402 moved $1.77M in 30 days, and 44% of it went to sellers that one wallet paid. Take those out and agents bought about $0.84M from 44 sellers, less than one Stripe startup, Stack Influence, collected on its own.

$10.13M

MRR of the 81 Stripe startups in TrustMRR's top 100; $5.29M from the 70 read in the last 30 days

$1.77M

paid over x402 in 30 days, in 4.72M payments from 30,541 buyers to 14,073 sellers

$0.84M

of that went to the 44 x402 sellers with ten or more buyers

Stripe: the top 100 on TrustMRR

TrustMRR ranks startups by MRR, and each startup’s page names the provider its revenue was read from. Stripe carries 81 of the top 100. RevenueCat and Superwall, which bill through the app stores, carry most of the rest, and those are mobile apps.

A figure on TrustMRR is the last one it read. When a founder’s key expires the page keeps showing it, so eleven of the 81 Stripe startups carry figures more than 30 days old. Stan, first on the board at $3.57M a month, was last read on 20 April. Without those eleven the Stripe column is $5.29M.

TrustMRR's top 100 by MRR, by payment provider; current means read in the last 30 days
ProviderStartupsMRRLast 30 daysCurrentCurrent MRR
Stripe81$10.13M$12.89M70$5.29M
RevenueCat12$658k$796k11$623k
Superwall4$221k$205k4$221k
Whop1$76k$186k1$76k
Paddle1$45k$48k1$45k
App Store1$26k$26k1$26k
Stripe startups whose figures were last read more than 30 days ago (the largest seven of eleven)
StartupMRR shownLast read
Stan$3,569,65420 April 2026
GojiberryAI$424,36814 August 2026
Kibu$234,31919 April 2026
A stealth company, US$135,89918 March 2026
PROSP$128,0055 February 2026
Supergrow$79,47928 February 2026
LLM Gateway$78,7647 September 2026

Who on Stripe collects the most

MRR counts subscriptions. What a startup collected in the last 30 days counts everything, and the four largest collectors take far more than their MRR: Stack Influence collected $872k against $28k of MRR, because brands pay it per campaign. Among the subscription businesses, Postiz grew 53% in a month to $444k, and 1Lookup, an API that checks phone numbers and addresses, collected $512k.

Stripe startups in the top 100 by money collected in the last 30 days, current figures only
StartupWhat it sellsLast 30 daysMRRChange, 30 days
Stack InfluenceInfluencer marketing$872,269$27,892+7%
Brand On Demand (Supliful)Products for creators' brands$825,193$177,622−16%
A stealth company, USEducation$633,550$171,207+3%
A stealth company, USMarketing$592,606$33,637+61%
1LookupPhone, email and IP checks$512,394$249,004+20%
PostizSocial media scheduling$443,690$403,954+53%
A stealth company, UKHealth$243,323$168,800−2%
A stealth companyNot stated$225,648$241,206+7%
CometlyAd attribution$221,393$196,525−18%
ReziResume builder$209,466$239,067−7%
AiSDRAI sales agent$189,832$194,312+1%
AEO EngineSearch and AI-answer SEO$173,542$188,713−24%

By category, AI tools hold the most Stripe MRR across the most startups: 22 of them, for $2.01M. Stan alone would put creator tools first, on figures from April.

Stripe MRR in TrustMRR’s top 100, by category (startups in brackets)

AI (22): $2,014,004AI (22)$2.01MSaaS (9): $769,178SaaS (9)$769kSocial media (6): $686,070Social media (6)$686kEducation (6): $594,421Education (6)$594kMarketing (9): $440,315Marketing (9)$440kAnalytics (5): $320,616Analytics (5)$321kE-commerce (3): $267,290E-commerce (3)$267k

x402: every payment on chain

An x402 payment is an HTTP request that carries its own payment: the server answers 402 with a price, the agent signs a USDC transfer, and a facilitator such as Coinbase settles it. x402scan sums them by window. The last 30 days are 3% of everything x402 has ever moved.

Every x402 payment x402scan indexes, by window; amounts in USDC
WindowVolumePaymentsBuyersSellers
24 hours$67,136105,0524,2391,835
7 days$439,0601,183,66410,7405,556
30 days$1,766,4444,717,10430,54114,073
All time$57,250,069238,695,279911,556321,970

The 100 largest sellers took $1.74M of the $1.77M, so ranking them accounts for nearly all the money. Many of them are not businesses. The largest received $719,000 in nine payments from one wallet, through FluxA, and a cluster of addresses on Meridian took about $10,000 each, from five to eight buyers apiece. Splitting volume by how many buyers paid each seller separates those from sales.

x402 volume of the top 100 sellers, last 30 days, by how many buyers paid each

One buyer: $782,387One buyer$782k2 to 9 buyers: $124,5942 to 9 buyers$125k10 or more buyers: $835,50510 or more buyers$836k

Base carried 84% of the volume and Solana the rest. Coinbase settled $850k, and FluxA $770k, almost all of it the one transfer.

Who on x402 is paid

A seller has a name on x402scan only if it lists itself in the x402 Bazaar. Of the 5,658 listed services, 15 took $1,000 or more in the last 30 days and 48 took $100. The largest named one is not an agent tool: laso.finance turns stablecoins into prepaid and gift cards. After it come speech and calls (dTelecom), research at two cents a call (AX1), and model gateways that charge under a cent.

x402 sellers by volume, last 30 days; named where listed in the x402 Bazaar
SellerVolumePaymentsBuyersWhat it is
0xfa02… (not listed)$719,00091One wallet's transfers through FluxA
0xc7c6… (not listed)$463,7613,185223Unknown; the largest seller that looks like commerce
laso.finance$150,720627113Prepaid and gift cards bought with stablecoins
7xvu… (Solana, not listed)$99,4504,330215Unknown
x402.dtelecom.org$36,67216,73529Speech-to-text, text-to-speech and calls
ax1.vc$31,0371,569,8239,407Venture research, about two cents a call
api-xpay.com$6,83260,907362Utility APIs
x402.clashofcoins.com$5,47936,817595Checkout for a game
sol.blockrun.ai$3,977617,813161116 models behind one endpoint, on Solana
korusmusic.co$3,4862,2271,135Music
api.botpay.network$1,60522,449367Payments for agents and machines
blockrun.ai$1,54995,436427Models and tool APIs, on Base
api.bitrefill.com$1,4281,851132Gift cards, eSIMs and top-ups
api.venice.ai$1,3421,10647Model inference
stableenrich.dev$1,26370,098292Lead enrichment and web data

The two side by side

The 81 Stripe startups in the top 100 collected $12.89M in the last 30 days, seven times all of x402, selling subscriptions and orders worth tens to thousands of dollars. x402 sellers charge per call: BlockRun about $0.006, AX1 about $0.02. The one x402 seller paid like a Stripe business, dTelecom, took $36,672 from 29 buyers, about $1,260 each.

What sells differs too. On Stripe it is AI software, marketing and education, sold to people. On x402 the money is in turning crypto into spendable money (laso.finance, Bitrefill) and in infrastructure sold to agents: speech, models, data. The same kinds of tools earn on Stripe already. 1Lookup, an API, collected more in a month than any x402 seller paid by more than one wallet.

Since our last reading

We read both sources on 14 October for what people and their agents pay for. x402 volume rose by nearly two thirds since, from fewer buyers and far fewer sellers, and laso.finance, then the largest seller, took about half as much.

The same figures, four weeks apart
14 October12 November
x402 volume, 30 days$1.09M$1.77M
x402 buyers, 30 days36,56330,541
x402 sellers, 30 days39,17514,073
laso.finance, 30 days$272,270$150,720
x402.dtelecom.org, 30 days$36,334$36,672
LLM Gateway, TrustMRR last 30 days$259,539$259,539, last read 7 September

Run it yourself

Every figure here came from the payments program in hedwigai’s shell. Add mcp.hedwigai.com to your MCP client, turn the program on once, and send these to its bash tool. Reading costs nothing.

$ programs enable payments
$ payments x402 stats 30                 # also 1, 7, or 0 for all time
$ payments x402 sellers 30               # by_buyers holds the one / 2-9 / 10+ split
$ payments x402 sellers 30 --listed true # Bazaar services, named
$ payments trustmrr growing stripe

TrustMRR’s ranked list needs an API key whose terms forbid republishing it, so the program reads each startup’s public page instead. These are the 100 slugs of the top 100 on 12 November 2026. Passed in one call, they give the provider table above in by_provider.

The command with all 100 slugs
$ payments trustmrr startup stan,superfruits,postiz,1lookup,kl23iod2mdka,anonymous-startup-25,rezi,kibu,stealth-venture-3,cometly,aisdr-inc,aeo-engine,brand-on-demand-inc,online-edtech,medical-practice-medical-consulting-agency,hidden-business-12,stealth-company-17,prosp,windmill-growth-llc,upscale-system-llc,propgpt-ai-props-analysis,unnamed-company-49,stealth-company-53,codedex,supergrow,llm-gateway,stealth-company-97,iker,ai-startup,promptwise,speel-co,podawaa,dropkiller,white-glove-content,project-a,omegleweb,stealth-02,unnamed-company-69,rankai,dm-champ,adspirer,traxy-ai,laper,boothie,dataexpert-techcreator,vidai-llc,post-bridge,conductor,uhoh-com,virtualjk,faceless-so,nutria-9e3eaa,kolbo-ai,rank-prompt,web3forms,backpedal-ltd,private-venture-84,anonymous-startup-2,indexsy-consulting-ltd,bazzly,dataford-inc,simple-analytics,mobile-app-4,audn-ai,chatseo,parkeur,apiframe,habitkit,launch-club,capgo,startup-532037cbd4f0,leadx,seobot,welya-inc,store-stock-checker,liinks,soundboost-ai-formerly-diktatorial,lunchbreak,reddit-agency,pushouse,ae-intelligence,based-labs-ai,stack-influence,tribe-social,datafast,convertlabs,stealth-03,produce-so,facekit,soldcomps,aikeyboard,stealth-company-16,sherpa,brawler,evergreen-support-llc,angel-match,mindset-app,walead-ai,monkei,verbi

The program calls these toolkit routes, which take the same fields as JSON. The schema is at toolkit.hedwigai.com/payments/openapi.json.

The payments program: each command, its toolkit route, and what it answers
CommandRouteAnswers
payments x402 stats 30POST /payments/x402/statsVolume, payments, buyers and sellers for 1, 7 or 30 days, or 0 for all time
payments x402 sellers 30POST /payments/x402/sellersEvery seller ranked, volume split by one, two to nine, and ten or more buyers, by chain and by facilitator
payments x402 sellers 30 --listed truePOST /payments/x402/sellersOnly the services listed in the x402 Bazaar, each named by its site
payments trustmrr startup <slugs>POST /payments/trustmrr/startupEach startup's MRR, last 30 days, all time, provider and sync date, summed by provider
payments trustmrr growing stripePOST /payments/trustmrr/growingTrustMRR's 25 fastest-growing and 25 newest startups

What these figures cannot say

TrustMRR is not a census. Founders choose to list, the largest companies on Stripe do not, and a page whose key expired keeps its last figures. The top 100 is ranked by MRR, so a startup that sells one-off orders can collect more than one on the list and still be missing from it.

x402scan counts every USDC transfer a known facilitator settles. A transfer between two wallets of one owner looks like a sale with one buyer, and one buyer with a thousand wallets looks like a market. The ten-or-more-buyers split is a filter, not a proof. Sellers that do not list themselves have no name, and two of the four largest are unnamed.

  • Everything was read on 12 November 2026.
  • Stripe and the other providers: the top 100 on TrustMRR by MRR, and each startup’s trustmrr.com/startup/<slug>.md page for its provider, figures and the date they were last read.
  • x402: x402scan, every seller over 30 days and totals over 1, 7 and 30 days and all time; amounts in USDC.
  • The earlier reading: What people and their agents pay for, 14 October 2026.

hedwigai

More head-to-head evals

GLM-5.3 against DeepSeek V4.1 Flash on 75 questionsTwo models answer the same 75 questions; one is always right, the other 14× cheaper.
MeasureGLM-5.3DeepSeek V4.1 Flash
Right, of 757571
Seconds a question, median3.91.8
Read the comparison
How we made a long-running AI agent for knowledge work 4× fasterOne cited knowledge work task through 13 experiments: 937 s to 233 s, with the model held constant.
MeasureFirst runBest since
Landscape, seconds937233
Memo, seconds759602
Read the report
Ten models plan a change to a spreadsheet: which do it in under a minute, traced call by callTen models drive one agent harness through the same spreadsheet request; six write the plan, timed and traced call by call.
MeasureQwen3 235BGLM-5.2Gemini 3.8 FlashClaude Opus 5
Seconds to the plan, median27485794
Plans written, of three runs2333
Read the comparison
SubQ, jev, Perplexity's Decider and Bespoke's Nimble on biotech diligence memos: finding the sentences a column is aboutFour decision models judge every sentence of ten biotech diligence memos for five columns, scored against sentences a person marked.
MeasureSubQjevDeciderNimble
Found, mean of five columns at section length64%65%66%68%
Mechanism of action, found at section length57%51%61%54%
Read the comparison
Leads with a reason to write: hedwigai ask, Apollo and treg on one brief40 leads each, every one checked: the right person, an inbox that accepts, and a dated reason to write.
MeasureaskApollotreg
Usable leads13 of 407 of 406 of 28
Cost per usable lead$0.037$0.27$0.23
Read the comparison
Self-hosted decisions for ask: Perplexity's open Decider in place of jev on 92 company questionsThe decisions inside hedwigai ask made by Perplexity's open Decider on two NVIDIA RTX 3090s, against jev.
Measureask with jevask with Decider
Questions right, of 928989
Wrong answers12
Read the comparison
Clinical decisions: jev and ask on System One, PubMedQA, SciFact and a 370-decision setEdgeEvals' System One board, PubMedQA, SciFact, and a clinical set built to System One's rules.
Measurejevask
Our 370 clinical decisions, right94.6%93.0%
PubMedQA, right (best EdgeEvals scored: 62.2%)77.0%74.4%
Read the comparison
Executive lookup for US-listed companies: Claude Opus 5.5 and ask on 95 questionsWho runs a public company, and how to email them: a frontier model from memory against hedwigai ask.
MeasureaskClaude Opus 5.5
Questions right, of 955963
Small companies right, of 20139
Read the comparison
Rental comps: floor-plan rents and amenities on 14 listings, five assistants comparedRent per floor plan and amenities for US rentals: hedwigai realestate-us against Claude, GPT, Gemini and Perplexity with web search.
Measurerealestate-usBest of four assistants
Floor-plan rents right, of 262623
Amenities claimed the listing doesn't state12129
Read the comparison
Company screening from SEC filings: jev, Vela 2.0 and ask on 92 yes/no questionsYes-or-no questions about public companies, answered by two decision models alone and by hedwigai ask.
Measureaskjev
Right, of the questions answered98%79%
Wrong answers, of 92219
Read the comparison
Defect localization in industrial images: four frontier models and a trained localizerFour frontier models asked to put a box around what they just described.
MeasureTrained localiserBest of four frontier models
Mean overlap with the defect (IoU, 0 to 1)0.4210.109
Read the comparison