Ant Group's blockchain arm unveils a platform enabling autonomous AI agents to hold assets, settle payments in USDC, and transact without human intervention — betting on a $3-5 trillion agent economy.
Ant Digital Technologies, the blockchain division of Chinese financial conglomerate Ant Group, has launched Anvita, a platform designed to let artificial intelligence agents autonomously hold digital assets, execute trades, and settle payments using stablecoins. Unveiled on 2 April at the company's Real Up summit in Cannes, Anvita represents one of the most ambitious attempts yet to build infrastructure for what its creators call the 'agent-to-agent economy' — a world in which software programs, not humans, are the primary participants in commercial transactions.
The launch places Ant Digital in direct competition with Visa, Google, Coinbase, and Mastercard, all of which have announced agent payment initiatives in recent months. But while most competitors are adapting existing payment rails for AI use, Anvita is built natively on blockchain infrastructure, integrating the open-source x402 protocol co-developed by Coinbase and Cloudflare to enable stablecoin payments directly over HTTP. The bet is that crypto-native rails will prove better suited to the speed, granularity, and programmability that autonomous agents require.
McKinsey has projected that AI agents could intermediate between $3 trillion and $5 trillion in global consumer commerce by 2030. Whether that forecast proves accurate or not, the race to build the payment infrastructure for autonomous software is accelerating rapidly, and Anvita's launch signals that the intersection of AI and blockchain is moving from theoretical discussion to commercial deployment.
How Anvita's Two-Layer Architecture Works
Anvita comprises two core products. The first, Anvita TaaS — tokenisation as a service — allows institutional clients to tokenise and manage real-world assets on blockchain. This layer extends Ant Digital's existing institutional blockchain business, which already supports tokenised assets from various financial institutions across Asia.
The second and more novel product is Anvita Flow, a platform where AI agents can register, discover other agents, collaborate on tasks, and settle payments in real time. Flow incorporates the x402 protocol, which repurposes the HTTP 402 'Payment Required' status code — a response type defined in the original HTTP specification but rarely used until now. When an AI agent requests a paid resource, the server returns a 402 response with payment instructions. The agent reads the instructions, signs a USDC transaction on-chain, attaches cryptographic proof of payment, and retries the request. The entire cycle completes in seconds without logins, subscriptions, or human approval.
Anvita Flow also includes an Agent Store, where developers can publish modules for data gathering, financial analytics, gaming, and other functions. The platform supports integration with modern AI development frameworks including OpenClaw and Anthropic's Claude Code, and offers flexible hosting options for agent deployment. According to Zhuoqun Bian, Ant Digital's head of blockchain, 'true transformation will come with the rise of fully autonomous software actively participating on-chain.'
The x402 Protocol and the Push to Make Payments Machine-Readable
The x402 protocol at the heart of Anvita Flow emerged from a recognition that existing payment systems were designed for humans, not software. Credit cards require identity verification, bank transfers involve batch processing delays, and even cryptocurrency payments typically assume a human is initiating the transaction. The x402 standard, by contrast, embeds payment logic directly into the HTTP request-response cycle, making financial transactions as programmatic as API calls.
Coinbase and Cloudflare released x402 as an open standard in late 2025, and adoption has been growing steadily if modestly. According to Artemis analytics, daily transaction volume on x402 rails currently averages approximately $28,000, though analysts flag that roughly half of observed transactions appear to be test or artificial activity rather than genuine commercial use. The protocol supports sub-cent micropayments — a capability that is impractical with traditional payment processors but essential for AI agents that may need to execute thousands of small transactions per hour.
The integration of x402 into Anvita gives Ant Digital a standards-based foundation rather than a proprietary lock-in, a strategic choice that may accelerate developer adoption. Solana, which has processed over 15 million agent-related transactions to date, has emerged as a primary chain for x402 deployments, though the protocol is chain-agnostic by design.
A Crowded Field: How Anvita Compares to Rival Agent Payment Systems
Ant Digital is entering an increasingly competitive landscape. Google unveiled its Agent Payments Protocol, or AP2, in September 2025, enabling AI agents to initiate payments through Google Pay infrastructure. Visa has been developing card-based agent payment systems that extend its existing network to autonomous software. And Mastercard's acquisition of blockchain payments firm BVNK in early 2026 signalled the card network's intention to compete on crypto-native rails as well as traditional ones.
What distinguishes Anvita is its integration depth with blockchain infrastructure. While Google's AP2 routes payments through conventional financial rails and Visa's system extends the card network model, Anvita is built entirely on-chain, settling transactions in USDC with finality measured in seconds rather than the days typical of traditional payment settlement. For AI agents operating in DeFi or other blockchain-native contexts, this eliminates an entire layer of friction.
Ant Digital also brings significant geographic reach. The company is currently pursuing stablecoin licences in Hong Kong, Singapore, and Luxembourg, and is working on USDC integration directly with Circle. Combined with Ant Group's existing Alipay+ network, which connects over 90 payment methods across Asia, Anvita could potentially bridge agent-to-agent crypto payments with the conventional consumer payments ecosystem — a capability none of its Western competitors currently offer.
From Prototype to Production: The Challenges Ahead
For all its ambition, Anvita faces significant headwinds. The agent-to-agent economy remains largely theoretical. The $28,000 in daily x402 volume — much of it synthetic — underscores the gap between the vision of autonomous AI commerce and its current reality. A survey by Gravitee found that only 14.4% of AI agents in enterprise environments have full security approval, while 88% of organisations report having deployed agents in some capacity, suggesting a mismatch between enthusiasm and operational readiness.
Regulatory uncertainty presents another obstacle. Ant Group's blockchain operations sit within a complex web of Chinese and international financial regulations, and the company's stablecoin ambitions will face scrutiny from regulators in every jurisdiction where it seeks licences. The memory of Ant Group's failed $37 billion IPO in November 2020, derailed by Chinese regulators over financial stability concerns, looms over any expansion into novel financial services.
Yet the underlying thesis — that software agents will eventually need their own payment infrastructure, and that blockchain provides the most natural foundation — has attracted serious capital and talent from across the technology and finance industries. Whether Anvita becomes the dominant platform or merely an early entrant in a market that will look very different by 2030, its launch marks a concrete step toward a future in which the most active participants in global commerce may not be human at all.
