Coinbase CEO Brian Armstrong has outlined a future in which artificial intelligence and cryptocurrency develop as interconnected trends rather than separate sectors. According to an original report, Armstrong said AI agents could eventually move more value each day than the entire human population, with those transactions running on blockchain infrastructure.

The statement comes as AI integration into financial systems continues to accelerate. Armstrong described a model in which autonomous agents make micropayments, pay service fees, and route capital without direct human intermediaries. Coinbase is positioning its own infrastructure stack — Base, USDC, and x402, a protocol designed for machine-to-machine payments — as a potential settlement layer for that agent-driven economy.

Coinbase has referred to this direction internally as “Agentic Finance.” The concept places the exchange at the intersection of artificial intelligence and on-chain financial infrastructure, with autonomous software agents able to interact economically through programmable blockchain-based payment rails.

Coinbase’s Infrastructure Bet

Armstrong’s comments are tied to infrastructure Coinbase is already developing. Base has become one of the more active Ethereum layer-2 networks, USDC maintains a major role in regulated stablecoin markets, and x402 is described as a technical specification that would allow AI agents to make small, frequent, verifiable payments without requiring a human wallet signature for every transaction.

The underlying argument is that traditional payment systems may not be suitable for a world in which large numbers of agents settle transactions in real time. Existing rails can involve settlement delays, chargeback risk, and high per-transaction costs. By contrast, a lower-cost layer-2 network with stablecoin support could function less as a standalone product and more as base-level infrastructure for autonomous payments.

Automated systems already account for substantial transaction activity in traditional financial markets, particularly through high-frequency trading. Armstrong’s view extends that automation beyond trading and into broader on-chain payments and smart contract interactions. In this model, AI agents would not only execute trades but also pay for compute resources, access APIs, settle invoices, and move funds across jurisdictions.

For those use cases, programmability and settlement finality become central requirements. Blockchain networks provide those features differently from legacy systems such as ACH or SWIFT, which were not designed for continuous machine-to-machine settlement at internet scale.

At the same time, the path to such a system remains uncertain. Crypto infrastructure continues to face challenges around scalability, user experience, and volatility management for routine payments. Even on Base, transaction costs can rise during periods of heavy activity, and stablecoin adoption remains concentrated in a limited number of jurisdictions. For an agent economy to reach the scale Armstrong described, on-chain throughput would need to increase substantially, and USDC would need broader use in the markets where agents operate.

The distance between a working technical prototype and global settlement rails for non-human counterparties remains significant. Coinbase’s view depends on the continued development of blockchain networks, stablecoin usage, and automated payment standards capable of supporting machine-driven transaction volume.

Regulation, Identity, and Liability

A major unresolved issue is identity. AI agents that conduct high-value transactions are likely to encounter the same anti-money laundering and know-your-customer requirements that apply to human users. A USDC-holding agent may need a way to demonstrate that it is not controlled by, or acting on behalf of, a sanctioned entity.

Coinbase’s regulated status may provide an advantage as these questions develop, but existing rules were not written for a machine-first financial environment. As lawmakers continue to debate crypto market structure, including recent Senate friction over major crypto legislation, the emergence of autonomous agents adds a further regulatory challenge that current frameworks only partially address.

Clarity on agent identity and liability could become as important as the underlying technology. If a payment made by an AI agent triggers a compliance flag, responsibility may be contested among the developer that deployed the agent, the user that funded it, and the protocol or platform that routed the transaction. These questions are not peripheral to Agentic Finance; they are part of the core design problem.

Coinbase’s strategy suggests that regulated entities able to manage these compliance and identity issues may be positioned to handle a meaningful share of agent-generated transaction activity if the market develops in that direction.

Coinbase’s Role in an Agent Economy

Armstrong’s framing also points to a broader strategic shift for Coinbase. Rather than treating AI only as a customer-service tool or as a component of trading bots, the company is presenting itself as a utility provider for autonomous economic activity. Under that model, Base would host agent contracts, USDC would serve as a unit of account, and x402 would operate as a payment gateway for machine-to-machine transactions.

Other projects are exploring related areas. UXLINK’s recent partnership with Origins Network reflects interest in scalable AI-driven Web3 applications, while on-chain activity around AI-related NFTs such as $X@AI shows continued demand for AI-themed crypto assets. Coinbase’s position combines exchange liquidity, stablecoin infrastructure, and a growing layer-2 ecosystem within a regulated framework.

Armstrong did not provide a date for when AI agent transactions could surpass human transactions. The infrastructure is still developing, and the agent economy remains at an early stage. However, Coinbase’s current direction indicates that the company is building beyond retail traders and institutions, with an emphasis on payment infrastructure for a machine-to-machine economy using crypto rails.