AI agent economy could process $3T-$5T by 2030 as crypto becomes payment infrastructure
Coinbase CEO Brian Armstrong said agent-economy payments could reach $3 trillion to $5 trillion by 2030, with crypto rails supporting real-time transactions between AI agents.

The AI agent economy could generate $3 trillion to $5 trillion in payments by 2030, with cryptocurrency emerging as a key financial infrastructure for transactions between agents.
In a CoinDesk interview, Coinbase CEO Brian Armstrong said AI should be decentralized alongside crypto, and that Coinbase is building financial accounts, payment rails and marketplaces for agent teams.
Armstrong said agents could handle more than 99.9% of all transactions within 10 years as their numbers surpass those of humans. Real-time, small-value payments between agents will require new rails, and Coinbase is testing such payments on Base using USDC stablecoins.
On security, Armstrong said users should segregate funds and set limits, controls and permissions for agents. Coinbase stores more than 99% of funds offline and uses AI to review and scan every line of code before production. Armstrong said he is running trading agents with $100,000 across several portfolios, with some up and some down.
On the recent crypto sell-off, Armstrong said Bitcoin may have already passed the low of its four-year cycle and could begin rising next year. He maintained his view that Bitcoin could reach $400,000 or even $1 million by 2030, citing regulatory clarity and growth in stablecoins, tokenized stocks, derivatives and prediction markets as positive signs. Prediction markets and derivatives now account for 25% of Coinbase's trading fee revenue, while tokenized stocks on Base have surpassed $4 billion to $5 billion.