BlackRock has published a research paper proposing a two-tier monetary system for autonomous AI agents. The firm argues that traditional banking infrastructure, including card networks and ACH transfers, is unsuitable for machine-to-machine payments due to settlement delays and the inability of software to open bank accounts independently.
The proposed model suggests using stablecoins for operational spending and Bitcoin as a store of value. BlackRock highlights that stablecoins offer near real-time settlement, which is essential for agents purchasing data or computing power. The company notes that stablecoin circulating supply exceeds $300 billion, with adjusted volume reaching $11 trillion in 2025 and $8.5 trillion in the first half of 2026.
The report references a February 2026 study by the Bitcoin Policy Institute, which analyzed 9,072 responses from 36 AI models. The models selected Bitcoin for value storage in 79.1% of cases and stablecoins for spending in 53.2% of instances. Traditional bank money was chosen in less than 9% of the surveyed scenarios.
BlackRock acknowledges that current payment volumes for AI agents remain small. The firm also notes that it is not advocating for a crypto-only ecosystem, as it recognizes competing payment solutions from companies like Stripe, OpenAI, Google, and Visa. BlackRock has already entered this market by launching a money market fund for stablecoin issuers.
