Visa, Mastercard, Coinbase, Stripe, Ant Group, and Google have each launched competing payment protocols for AI agents in the past 90 days, collectively positioning for what McKinsey projects could be a $3 trillion to $5 trillion agentic commerce market by 2030. The infrastructure buildout is rea...
"I haven't seen anything like this since the dawn of ecommerce itself in the late '90s or early 2000s." — Jack Forestell, Chief Product & Strategy Officer, Visa
Visa, Mastercard, Coinbase, Stripe, Ant Group, and Google have each launched competing payment protocols for AI agents in the past 90 days, collectively positioning for what McKinsey projects could be a $3 trillion to $5 trillion agentic commerce market by 2030. The infrastructure buildout is real: Visa's Trusted Agent Protocol, Stripe's Machine Payments Protocol, Coinbase's x402 standard, and Ant Group's Anvita platform all went live in Q1 2026.
The adoption is not. Coinbase's x402 protocol — the most mature on-chain standard — processes approximately $28,000 in daily volume, according to Artemis on-chain data. Much of that activity consists of self-dealing and wash transactions. Mastercard paid $1.8 billion for stablecoin startup BVNK. Solana reports 15 million cumulative AI agent transactions. But fewer than 1% of consumers use agents to purchase anything, per Morgan Stanley.
The gap between infrastructure investment and actual throughput defines the agentic payments sector in April 2026. Every major payments network is building for a future that remains, by the data, almost entirely theoretical.
Six competing protocols now claim to define how AI agents will pay for goods and services:
| Protocol | Sponsor | Rail Type | Status | |----------|---------|-----------|--------| | Trusted Agent Protocol | Visa / Cloudflare | Card + crypto | Live (Q1 2026) | | Machine Payments Protocol | Stripe / Tempo | Card + stablecoin | Live (Q1 2026) | | x402 | Coinbase / Cloudflare | On-chain (USDC) | Live (Linux Foundation) | | Anvita Flow | Ant Digital Technologies | On-chain (USDC) | Live (March 31, 2026) | | Agent Commerce Protocol | OpenAI | Card-based | Live (Q1 2026) | | Universal Commerce Protocol | Google | Multi-rail | In development (60+ org consortium) |
No interoperability standard exists between these protocols. Visa has positioned itself as a meta-layer, announcing support for four of the six standards through its Intelligent Commerce platform. Stripe claims to be the first provider supporting both agentic network tokens and buy-now-pay-later tokens through a single API primitive.
The fragmentation echoes early mobile payments circa 2012, when Apple Pay, Google Wallet, CurrentC, and carrier-based solutions competed before consolidation. The difference: the agentic payments stack must handle sub-cent micropayments at machine speed — a use case that traditional card authorization flows were not designed for.
Coinbase's x402 protocol, which revives the HTTP 402 "Payment Required" status code for machine-to-machine payments, has processed over 50 million transactions since launch and supports approximately 500,000 active AI wallets, according to Coinbase. The protocol transitioned to the Linux Foundation as a vendor-neutral standard.
The headline numbers obscure a demand problem. Artemis, the on-chain analytics firm, found that x402's actual daily volume hovers around $28,000, generated from roughly 131,000 daily transactions at an average payment of $0.20. Artemis categorizes a significant share of these transactions as "gamed" — either self-dealing, where the same wallet acts as both buyer and seller, or wash trading, where the seller funds the buyer's wallet and the funds return immediately.
This places x402's real commercial throughput at a fraction of its reported metrics, despite an ecosystem valuation that CoinDesk estimated at approximately $7 billion.
Solana accounts for approximately 65% of x402's on-chain payment volume, reflecting the network's sub-cent fees and high throughput. Base, Coinbase's Ethereum L2, handles most of the remainder. Coinbase initially rolled out Agentic Wallets on Base in February 2026, with Solana, Polygon, and Arbitrum support planned later in the year.
Ant Digital Technologies, the blockchain division of Alipay parent Ant Group, unveiled Anvita on March 31, 2026, at the Real Up summit in Cannes. The platform consists of two products:
Anvita TaaS (Tokenization-as-a-Service): Institutional-grade real-world asset tokenization with custody and treasury management tools. The product targets banks and asset managers seeking to digitize traditional securities.
Anvita Flow: A network enabling AI agents to register, discover counterparties, coordinate multi-step tasks, and settle payments in real time using USDC via the x402 protocol.
"Pure asset tokenization is only the static infrastructure," said Zhuoqun Bian, Ant Digital Technologies' blockchain president. The transformation, according to Bian, lies in "on-chain agentic economies where autonomous systems actively manage assets and portfolios independently."
Anvita Flow is compatible with Claude Code, OpenAI, and standard agent toolkits. The platform is seeking stablecoin licenses in Hong Kong, Singapore, and Luxembourg — a regulatory prerequisite that most crypto-native competitors have not yet secured.
The institutional framing is deliberate. Where Coinbase targets developer-first adoption and Visa targets merchant integration, Ant Group is building for regulated financial institutions that require custody, compliance, and multi-jurisdictional licensing before deploying autonomous agents.
The two largest payment networks are approaching agentic commerce from opposite directions.
Visa has focused on protocol proliferation. Its Intelligent Commerce platform, launched in Q1 2026, supports Visa's own Trusted Agent Protocol, Stripe's Machine Payments Protocol, OpenAI's Agent Commerce Protocol, and Google's Universal Commerce Protocol. Visa partnered with Akamai Technologies and AWS for fraud prevention infrastructure. The strategy: become the neutral authentication and settlement layer regardless of which protocol wins.
Rubail Birwadker, Visa's global head of growth products and strategic partnerships, stated that agents need identity verification, data collection, and security at every transaction step. Birwadker characterized the potential as "next level one-click checkout, perhaps even without a button."
Mastercard has focused on acquisition. The company agreed on March 17, 2026 to acquire stablecoin infrastructure startup BVNK for up to $1.8 billion, including $300 million in performance-contingent payments. According to Bloomberg, it is Mastercard's largest cryptocurrency deal to date. BVNK provides technology to bridge fiat and blockchain-based transactions across more than 130 countries.
Mastercard CEO Michael Miebach said the network completed its "first agentic transaction" in Q3 2025 and declared that Mastercard would be "at the center" of agentic commerce. The company launched Agent Pay and a broader agent suite in January 2026, followed by virtual AI-powered business management tools for SMBs in March 2026.
The combined signal: card networks view agentic payments not as a crypto sideshow but as the next major transaction category after e-commerce and mobile.
The Solana Foundation disclosed at the Blockworks Digital Asset Summit that the network has processed over 15 million on-chain AI agent transactions in 2026, with stablecoins as the default settlement mechanism.
Vibhu Norby, Chief Product Officer at Solana Foundation, predicted that "99.99% of on-chain transactions" would be initiated by agents, bots, or LLM-based wallets within two years. The claim is directionally plausible — machine-generated transactions already dominate Solana's throughput via MEV bots and automated trading — but conflates different types of programmatic activity.
Solana's appeal for agentic payments is structural. Transaction costs below $0.001 and finality under 500 milliseconds make the network viable for sub-cent micropayments. By contrast, Ethereum L1 transaction costs, even post-Dencun, remain prohibitive for per-API-call billing. Ethereum L2s have narrowed the gap — costs have dropped from $24 to under one cent, a 2,400x decrease — but Solana's throughput of 3,400 TPS provides headroom that L2s have not yet matched under sustained load.
The Foundation launched an enterprise platform with Mastercard and positioned Solana as core infrastructure for what it calls the "agentic internet." Whether 15 million transactions represent meaningful economic activity or primarily testing and experimentation remains an open question. The Foundation has not disclosed the dollar value of these transactions.
The fundamental tension in agentic payments is that the use case demands near-zero fees on transactions worth fractions of a cent, while existing revenue models in payments depend on percentage-based takes of larger transactions.
Stablecoin transaction volume reached $46 trillion annually as of early 2026, up 106% year-over-year, according to industry data. But real-world payments represent only 1% of the $26 trillion in on-chain stablecoin volume — the rest is trading, DeFi activity, and transfers. Adjusted for automated trading, stablecoin volume was $9 trillion, still up 87% year-over-year.
The agentic use case introduces a new category: high-frequency, low-value, machine-initiated payments. An AI agent calling an API 10,000 times per day at $0.001 per call generates $10 in daily volume. Scaled across 500,000 active wallets, this produces $5 million daily — still a rounding error for networks processing trillions.
For the $3 trillion to $5 trillion McKinsey projection to materialize, agent-initiated transactions must move beyond API micropayments into larger commerce categories: procurement, logistics, and financial services. Morgan Stanley's finding that only 1% of consumers currently use agents for purchasing suggests this transition is years away. A survey by Deloitte found 87% of executives cite trust as the primary adoption barrier.
Enterprise adoption projections from Gartner suggest agentic AI deployment will grow from less than 1% of enterprises in 2024 to 33% by 2028. But "deployment" and "autonomous purchasing authority" are different thresholds. McKinsey's own research identified that 80% of organizations have observed risky AI agent behaviors, including unauthorized data exposure and privilege escalation.
The agentic payments infrastructure buildout of Q1 2026 represents one of the largest pre-revenue bets in payments history. Mastercard alone committed $1.8 billion. Visa rebuilt its product stack. Coinbase migrated its core protocol to a foundation. Ant Group launched a two-product suite targeting regulated institutions.
The gap between investment and adoption is not unusual for infrastructure cycles — early internet payment rails preceded e-commerce volume by years. What distinguishes the current cycle is the speed of protocol proliferation without corresponding demand. Six standards competing for a market generating $28,000 in daily on-chain volume is a ratio that favors consolidation.
The economic value will ultimately flow to whichever protocol achieves two things: merchant integration at scale and regulatory clarity across jurisdictions. Visa's multi-protocol strategy hedges broadly. Ant Group's licensing-first approach targets institutional trust. Coinbase's open-source bet assumes developer adoption drives network effects.
None has yet demonstrated that AI agents, in their current form, generate sufficient economic activity to justify the infrastructure being built for them. The data suggests the market is 2-4 years ahead of itself. That does not mean the thesis is wrong. It means the thesis is early.