Notice: Analysis based on data consulted on August 3, 2026 (x402.org public dashboard and PING price via CoinGecko API) and time series from Chainalysis, Artemis via CoinDesk, Dune, the Linux Foundation, Cloudflare, and the Solana Foundation, each with its respective cutoff date indicated in the text. Live dashboard metrics change week by week. Nothing that follows constitutes financial advice or an opinion on the future price of any token. CleanSky does not receive commissions or referral payments from Coinbase, Solana, Base, Polygon, Cloudflare, or any project mentioned.
PING, the token that half the market adopted as a thermometer for x402, was trading around $0.00116 on August 3, 2026: 98.5% below its all-time high on October 25, 2025, with approximately $31,700 in 24-hour volume. The protocol attributed to this collapse does not emit any token: x402 settles in USDC—Circle's stablecoin, a tokenized dollar—and its official dashboard showed 75.41 million transactions and $24.24 million moved in the last 30 days on that same day. In between, there are three data series that coverage tends to merge into a single headline: the raw transaction count, which plummeted more than 90% between December 2025 and February 2026; the composition of those payments, which improved over the same period until 95% of the volume traveled in transactions above one dollar; and the price of speculative tokens that were never part of the standard. This article separates the three, clarifies exactly what each dashboard measures—because they do not measure the same thing—and adds a first-hand measurement on the other half of the matter: what happens when a site lets AI crawlers in for free instead of charging them.
Is x402 real or just hype?
The name comes from the HTTP 402 "Payment Required" code, a status the web reserved in the nineties and left dormant for three decades until Coinbase woke it up in May 2025. The full mechanism—the server responds with 402 and payment instructions, the agent pays in stablecoin and repeats the request with the proof—was broken down in our analysis of the permission model for AI agents in DeFi, and the comparison with rival standards from Google and Stripe can be found in the article on privacy stablecoins. What matters here is what the numbers say a year and three months later.
The answer to the question "is it real or hype?" changes depending on which series you look at, and that ambiguity is the trap. The daily count since the peak depicts a disaster; the volume composition shows an orderly maturation; the PING chart looks like a closed casino. All three images are true and describe different phenomena: the inflated activity that evaporated, the real payment activity that stayed, and the speculation built on top of the name without touching the standard.
How much did the raw x402 transaction count plummet?
The peak is well-documented. The daily average went from 731,000 transactions in December 2025 to about 57,000 in February 2026: a 92% decrease in the same aggregate and using the same methodology. Specific spikes were even higher: a single day in February recorded 3.8 million transactions with approximately $2 million moved, according to data from on-chain analytics firm Artemis reported by CoinDesk.
Solana offers the cleanest series because it is the most granular. In the week of December 22, 2025, the network processed 6.8 million x402 transactions. In the week of February 9, 2026, it fell below 510,000—more than 90% less in seven weeks. By March 11, 2026, when CoinDesk published its review, the protocol was moving around $28,000 daily spread across approximately 131,000 transactions: an average ticket of $0.21.
The uncomfortable part of the peak is that a large portion was not real demand. Artemis estimated that around 50% of transactions during the height were "gamified"—operations between wallets of the same operator to inflate the count, the equivalent of wash trading in an order book. An earlier measurement by the same firm in late 2025 marked 48% of the count and 81% of the dollar volume as non-organic, and its analyst described the boom in agentic payments as "still mostly a mirage." Once that filter is applied, more than half of the peak was never demand, and the subsequent drop primarily removes that self-generated activity.
Why is payment quality improving while the count falls?
Chainalysis published a breakdown on June 3, 2026, that allows for cross-referencing. At the beginning of 2025, transactions over one dollar accounted for 49% of the settled volume, while the $0.10 to $1 range took 46%; a year later, the former reached 95% and the latter had dropped to 4%. The conversion from one-time trial users to recurring payers quadrupled in six months, and weekly retention, volatile between 5% and 87%, maintained an upward trend after the speculative trigger disappeared. The full table is in our analysis of agentic permissions.
The profile of those paying reinforces this reading. Wallets executing x402 payments have an average age of 197 days compared to 423 for the typical Base wallet, hold about 26 different tokens compared to the average of 4, and received capital inflows about twelve times larger: young, active, and well-funded accounts. When placed by date alongside the activity series, the two diverge clearly.
| Cutoff Date | Raw Activity Count | Payment Composition and Quality | Source / Perimeter |
|---|---|---|---|
| Early 2025 | Base starts practically from zero | 49% of volume in payments >$1; 46% between $0.10 and $1 | Chainalysis (Base only) |
| Nov-Dec 2025 (Peak) | Avg 731,000 tx/day in Dec; Solana 6.8M tx in week of Dec 22 | 48% of count and 81% of volume marked as non-organic (late 2025) | Artemis via CoinDesk (multi-chain aggregate) |
| Week of Feb 9, 2026 | Solana below 510,000 tx/week (−92% vs. December) | — | Dune via SolanaFloor (Solana only, weekly) |
| Mar 11, 2026 | ~131,000 tx/day and ~$28,000/day; avg ticket $0.21 | ~50% of peak tx considered gamified | Artemis via CoinDesk (aggregate) |
| Apr 14, 2026 | 120M cumulative tx and $41M settled; 57,000-90,000 tx/day | Cumulative avg ticket ~$0.05 | Dune via Web3Trackers (14 chains, historical) |
| Jun 3, 2026 | >100M cumulative tx on Base | 95% of volume in payments >$1; 4% between $0.10 and $1; 4x recurring payer conversion | Chainalysis (Base only, cumulative) |
| Aug 3, 2026 | 75.41M tx; 94,060 buyers; 22,000 sellers | $24.24M settled; avg ticket $0.32 | x402.org Dashboard (30-day rolling, all chains) |
The last row resolves the apparent contradiction between columns: an average ticket of $0.32 coexists perfectly with "95% of volume travels in payments of more than one dollar," because one counts operations and the other distributes dollars. Millions of cent-sized micropayments survive, but the settled dollars are concentrated in large payments. Furthermore, that average ticket has increased sixfold from the $0.05 cumulative average in April.
What happened to PING and the "x402 ecosystem" tokens?
PING was launched in October 2025 on Base, and CoinGecko classifies it as a memecoin within the "x402 Ecosystem" category. It hit its all-time high of $0.078576 on October 25, 2025, reaching a market cap of over $80 million. The query on August 3, 2026, returns approximately $0.00116, a $1.16 million market cap, and about $31,700 in 24-hour volume: 98.5% below its peak.
The speculative sector as a whole followed the same path. The aggregate DEX trading volume for these tokens touched about $970,000 daily on December 3, 2025; DefiLlama recorded $572,000 over 30 days in mid-July 2026, which is roughly $19,000 per day. Two orders of magnitude less in seven months.
The point that changes the headline's reading is simple to verify but hard to remember when looking at a red chart: x402 has no token. It is an open standard over HTTP governed by the Linux Foundation that settles in stablecoins, almost always USDC. Buying PING provided as much exposure to the protocol as buying a token named "HTTPS" would to the encrypted web. That collapse does, however, explain part of the 87% retention peak recorded by Chainalysis and its subsequent drop to 5%: when the incentive was turned off, those who came for it left.
Which chain captures the volume today: Solana, Base, or Polygon?
It depends on what is being counted and in what window, and that nuance makes this section a minefield for anyone quoting a single figure. Base dominates the historical cumulative total, Solana holds the recent monthly share, and Polygon was the one that broke Solana's monopoly in February. Furthermore, the aggregate published by x402.org is up to forty times higher than what can be deduced from Dune dashboards per chain for the same months.
| Series | What it exactly measures | Figure | Cutoff Date |
|---|---|---|---|
| Coinbase — "Year 1" | Cumulative payments since launch, all chains | 169M payments; 590,000 buyers; >100,000 sellers | May 2026 |
| Dune via Web3Trackers | Cumulative transactions and USDC settled, 14 chains | Base ~70M tx / $21.5M; Solana ~45M tx / $16.4M; others ~5M tx / $3.1M | Apr 14, 2026 |
| Dune, Solana weekly share | % of x402 transactions on Solana over total, per week | 88% (week of Feb 2) → 49.7% weeks later | Feb 2026 |
| Base Team | Transactions and value on Base in 30 days | 3.1M tx; $1.2M; +37% buyers; +23% sellers | May 29, 2026 |
| Solana Foundation (KSNet agreement note) | Monthly transaction volume share attributed to Solana | 70% of monthly volume; >37M tx; >20,000 buyers and sellers | Jul 30, 2026 |
| Official x402.org Dashboard | Transactions and dollars settled, 30-day rolling, all chains | 75.41M tx; $24.24M; 94,060 buyers; 22,000 sellers | Aug 3, 2026 reading |
| XRP Ledger | Autonomous agent transactions since June 9 integration | 1,434,517 tx (+127%); 129 merchants settled 4,463.24 XRP and 1,895.14 RLUSD | Jul 22, 2026 |
Two reconciliations are worth noting. First: the 37 million transactions the Solana Foundation attributed to its network on July 30 represent 49% of the 75.41 million on the aggregate dashboard, so the "70% of monthly volume" in that same note must be measuring settled dollars rather than operations, or a different window. Both figures describe Solana as the monthly leader, but they are not interchangeable.
The second is more significant: 75.41 million transactions in 30 days equates to about 2.5 million daily, a pace comparable to the November 2025 peak and between twenty-eight and forty-four times higher than the 57,000-90,000 daily shown by Dune between February and April 2026. Either activity re-accelerated sharply in the second quarter, or the official dashboard counts events that per-chain dashboards do not see—such as 402 responses completed outside of indexed facilitators. No public source currently bridges this gap: the dashboard does not publish its methodology. What does withstand contrast is comparing within the same series: Solana weekly against Solana weekly, Base against Base.
Why are Visa, Mastercard, and Cloudflare entering right now?
While activity series went through the full cycle of surge and purge, the institutional layer moved in the opposite direction without pause. Milestones from the last month and a half:
- July 1, 2026 — Cloudflare launches the Monetization Gateway. Any resource protected by Cloudflare (pages, APIs, MCP datasets, tools) can be charged via x402 with configurable rules per request type, and settlement occurs in stablecoins. On the same day, it announces that its Pay Per Crawl marketplace is evolving into Pay Per Use, with Ceramic.ai and You.com as initial partners.
- July 14, 2026 — the x402 Foundation becomes operational under the Linux Foundation with 40 organizations. Premier members include Visa, Mastercard, American Express, Stripe, Adyen, Fiserv, Shopify, Google, Amazon Web Services, Cloudflare, Circle, Coinbase, MoonPay, Ripple, the Stellar Development Foundation, the Monad Foundation, and the Solana Foundation. General members include Polygon Labs, Fireblocks, Injective, Quant Network, LayerZero Labs, NEAR Foundation, KakaoPay, and World Liberty Financial, among others.
- July 22, 2026 — XRP Ledger reports 1,434,517 autonomous agent transactions since its June 9 integration, up 127%, with 129 merchants settling in XRP and RLUSD.
- July 23, 2026 — Coinbase presents its Agent Payment Stack for merchants to accept agent payments, alongside "Coinbase for Agents" and a third-party SDK. Its business unit closed the first year with about 5,000 clients and around $1 billion between payments and trading. The highlight from its lead: since June, 53% of web traffic to Base documentation comes from agentic visitors.
- July 30, 2026 — the Solana Foundation signs a memorandum of understanding with KSNet, a Korean payment operator with a 26-year track record, to pilot Solana Pay and later x402 with Korean merchants, respecting anti-money laundering frameworks and settlement in won.
The pattern in this list: none of these actors are betting on the price of anything. They are integrating a charging standard into infrastructure they already have deployed, in the same quarter that speculative counts were collapsing. For Visa or Adyen, the appeal is the part that cards never solved—a cent-sized charge with a sub-cent fee and settlement in about 200 milliseconds, compared to $0.30 plus 2.9% for a classic gateway, a breakdown we covered when analyzing Base as Coinbase's invisible layer. Cloudflare pursues the same from the other end: turning crawling traffic that it currently serves for free into billable revenue.
Who charges for AI crawling and who lets it pass for free?
During its Agents Week in April 2026, Cloudflare estimated that the HTTP 402 responses its network sites return to AI crawlers—programs that automatically traverse pages to copy content—exceed one billion daily. It is, by far, the largest real use of the 402 code today, well above payments between agents.
The other side of that decision can be measured precisely on a small site. Between July 1 and July 28, 2026, cleansky.io access logs recorded 28,617 requests from identified AI crawlers, all served without any payment barrier:
- 12,677 requests with training intent (44%) — copies intended to feed future models; a single crawler accounted for 7,930 of them.
- 9,884 indexing requests (35%) — building search indices for assistants, with one dominant crawler totaling 7,655.
- 6,056 live query requests (21%) — the crawler searching the page at the moment an assistant responds to a user.
- 40 human visits referred from AI assistants in those same 28 days.
That last line provides the proportion: 715 crawl requests for every one human visit returned. Cloudflare describes the range of the phenomenon as "from one hundred to tens of thousands of requests for every visitor they return"; our own measurement falls right within that bracket and puts a concrete number on it for a real site. At a hypothetical price of one cent per request—each publisher sets their own, there is no market rate—that month would have generated about $286 in revenue.
The comparison with Cloudflare is one of model, not scale: a network that serves a substantial portion of global traffic is not measured in magnitude against an independent site. What is comparable is the decision. A publisher who lives off advertising loses money every time a crawler copies their page without returning visits, and for them, charging is elementary arithmetic. A site whose goal is to be cited as a source in the responses generated by those systems has the opposite incentive: charging for entry is equivalent to paying to disappear.
What to watch before September 15, 2026?
This is the next inflection point on the calendar. Starting September 15, 2026, Cloudflare will block training and agent crawlers by default on pages with advertising, and will completely block "mixed-use" crawlers—those that do not declare whether a specific request is for search indexing or copying for training. This will apply by default to all free plan customers and any new sign-ups after that date; existing paid customers will be able to override it from their dashboard. The measure targets bots that currently bundle search indexing with training and inference.
If a large part of the web closes its doors except for payment, the demand for a standardized mechanism to open them no longer depends on agents buying APIs of their own volition. Three indicators allow for tracking this without waiting for headlines:
- The average ticket on the x402.org dashboard. It went from $0.05 in the April cumulative total to $0.32 in the August 3 reading. If it continues to rise, the weight is being driven by real service payments (inference, market data, browser sessions) rather than trial micropayments.
- Weekly share per chain, compared to itself. Solana went from 88% in February to half the count with Polygon's entry, while Base maintains the historical cumulative lead. What matters is which network retains recurring buyers, not who wins a single week; we discussed the context of Solana's payment infrastructure when analyzing Jupiter as a superapp.
- The number of sellers. This is the hardest figure to inflate on the entire dashboard: 22,000 in the August 3 reading, compared to the over 100,000 cumulative sellers reported by Coinbase in its first year. An agent can manufacture transactions with itself; a catalog of services that actually charge takes much longer to appear.
With the standard governed by a foundation including Visa, Mastercard, and AWS, the risk of x402 disappearing is low. The unknown is whether 2027 volume will come from autonomous agents buying services—the original promise—or from publishers charging tolls to the crawlers of large AI platforms, the use case currently generating a billion 402 responses daily. Both use the same HTTP code and the same stablecoins as a settlement rail, yet they describe very different economies.
Related articles: The permission model for AI agents in DeFi. Base: Coinbase's invisible L2. Privacy stablecoins and rival payment standards. Stablecoins as a payment rail: 321 billion and their risks. Track your positions and stablecoins on-chain at CleanSky — no yield promises, just your data.