Key Highlights
- Armstrong argued that the number of AI agents making payments may eventually exceed the total number of people.
- According to him, traditional banking systems present a barrier for software-based entities because they cannot open accounts.
Coinbase CEO Brian Armstrong says autonomous artificial intelligence systems could soon conduct more transactions than human users.
In a post on X on Monday, Armstrong argued that the number of AI agents making payments may eventually exceed the total number of people using digital payment systems.
According to him, traditional banking systems present a barrier for software-based entities because they cannot open accounts. However, crypto wallets can be created using private keys and do not require the same identity verification procedures.
“Very soon, there are going to be more AI agents than humans making transactions,” Armstrong wrote, adding that while machines cannot hold bank accounts, they can operate cryptocurrency wallets.
Crypto wallets as a payment layer for AI
Armstrong’s comments came after Coinbase introduced Agentic Wallets in February 2026. The product allows automated software agents to hold and transfer digital assets.
By integrating gasless transactions and programmable spending limits on the Base network, the system lets software agents manage funds, hold identity, and execute transactions without human intervention.
Industry figures highlight machine payments
Others in the crypto industry have made similar claims about the potential scale of AI-driven transactions.
Changpeng Zhao (CZ), founder of Binance, recently suggested that autonomous agents could generate far more payments than human users if the technology becomes widely deployed.
In an X post on Monday, Zhao wrote that AI systems may eventually make orders of magnitude more transactions, particularly if automated software begins performing routine digital payments.
The idea of an “agent-native” financial system
The discussion reflects a broader debate about whether the current financial infrastructure can support autonomous software.
Banks, payment processors, and legal frameworks are generally designed for individuals or organizations with legal identities. Machines cannot easily participate in these systems because they cannot hold identification documents or sign legal agreements.
As AI agents gain the ability to make economic decisions, some developers argue they will require a form of “agent-native” money, a system that allows software to send and receive payments without human intermediaries.
Also Read: US Banking Giants Eye Lawsuit as OCC Opens Door to Crypto Firms
