BlackRock Proposes a Two‑Tier Money System for AI Agents

BlackRock released a research paper outlining a monetary framework tailored for autonomous AI agents. The paper claims that traditional payment rails such as card networks and ACH transfers are ill‑suited for software that must pay other software instantly.

Why Existing Payment Methods Fall Short

According to the paper, an AI agent cannot open a bank account or obtain a payment card without human involvement. Merchant fees make sub‑dollar transactions uneconomical, and ACH transfers can take up to a business day to settle. In contrast, blockchain networks settle transactions around the clock in near real time.

Two‑Tier Model: Stablecoins for Spending, Bitcoin for Saving

The report proposes a two‑tier system. Stablecoins, described as “machine‑native money,” would handle day‑to‑day spending because they settle quickly and are already widely used in the crypto ecosystem. For longer‑term value storage, the paper cites a study by the Bitcoin Policy Institute (BPI) indicating that a majority of AI models preferred Bitcoin as a store of value.

Supporting Data

• BlackRock notes that the circulating supply of stablecoins exceeds $300 billion.
• Adjusted stablecoin transaction volume reached $11 trillion in 2025, comparable to Visa’s annual volume.
• Stablecoins processed an additional $8.5 trillion in the first half of 2026.

AI Model Preferences

The BPI study surveyed 36 frontier AI models from Anthropic, OpenAI, Google, xAI and DeepSeek, collecting 9,072 responses. When asked where to store value, 79.1 % of answers favored Bitcoin; when asked what to spend, 53.2 % favored stablecoins. Bank money received less than 9 % of the total preference.

Implications and Caveats

BlackRock emphasizes that its conclusions are speculative and not financial advice. The paper also acknowledges that live payment volume from AI agents remains small and that other payment rails, such as those from Stripe, OpenAI, Google and Visa, continue to operate with traditional fiat.