THE DAY, IN CONTEXTPolaris InsightsAn AI-powered comprehensive news portal

POLARIS INSIGHTS

AI Agents Are Already Testing Blockchain Payments, Giving Coinbase and Robinhood Different Stakes in the Emerging Market

AI Agents Are Already Testing Blockchain Payments, Giving Coinbase and Robinhood Different Stakes in the Emerging Market

Artificial-intelligence agents are beginning to pay for data, software, computing and other services through blockchain-based systems, offering an early test of whether autonomous software can become a meaningful source of crypto activity. The x402 protocol, developed by Coinbase, enables stablecoin payments within web requests without requiring a conventional account or subscription.

The activity has drawn attention from major investors. Andreessen Horowitz’s crypto investment group previously described a system in which agents would use cryptographically signed credentials to transact with one another and pay for data or computing resources. BlackRock has likewise identified stablecoins as a likely early beneficiary of agent-driven commerce because they can support automated, small-value payments, while noting that standardized and liquid markets for computing capacity are not yet developed.

Coinbase has exposure at several points in that system. Its Base network hosted much of the reported x402 activity, and Coinbase has integrated the protocol into tools that let agents pay for information, model inference and other services. Base has also promoted agent wallets, spending controls and services that allow agents to pay in USDC for search, inference, browser access and other functions.

The financial payoff for Coinbase is not guaranteed by transaction growth alone. Low fees on Base can limit direct network revenue, while the company’s larger potential exposure comes from USDC-related income and its role in facilitating payments. Coinbase has also expanded its agent products beyond crypto: its agent tools now support U.S. Stocks and exchange-traded funds, allowing an agent to research an asset, pay for additional data and place a trade within user-defined limits.

Robinhood is pursuing a different route. Robinhood Chain, an Ethereum Layer 2 built with Arbitrum technology, went live on July 1 with infrastructure for tokenized assets and onchain financial products. The company has also launched agentic trading, allowing customers to connect an outside AI agent to a separate account with limited funds, notifications, activity monitoring and the ability to disconnect the agent. Its agentic credit-card product similarly uses dedicated virtual cards, spending limits and optional manual approvals.

That gives Robinhood a direct way to put agents inside its brokerage and payments ecosystem, but it does not provide the same reported connection to stablecoin-reserve income that Coinbase has. Robinhood’s agent business therefore remains more dependent on adoption of its broader brokerage, card and blockchain products. The company has warned that agents can misread instructions, act on incomplete information or make unexpected decisions, leaving customers responsible for monitoring activity and losses.

The emerging split is clear: Coinbase is more directly tied to the payment and settlement layer through Base, x402 and USDC, while Robinhood is using agents as a new interface for trading and spending. Both strategies depend on whether automated software develops from early experiments into sustained economic activity; the verified data so far shows rising infrastructure and usage, but not yet a mature agent economy.

Read the full account of the story

More from the AI desk

More from AI · Back to Polaris Insights