BlackRock says AI agents need machine-native money

BlackRock says AI agents need machine-native money. The paper identifies stablecoins as the leading candidate for machine spending. It proposes Bitcoin for saving. The paper says card networks and bank transfers were not built for software that pays software. An autonomous agent cannot open a bank account or obtain a card without human involvement. Merchant fees can make sub-dollar payments impractical. ACH transfers can take up to a business day to settle. The paper says blockchains settle around the clock in near real time. BlackRock says stablecoins, native cryptoassets, and other on-chain assets can serve as machine-native instruments. The paper cites Coinbase's x402 protocol as an example of agents paying for data on the spot. The paper also lists payment systems from Stripe, OpenAI, Google, and Visa. Some of those systems settle in ordinary bank money. The paper says live agent payment volume remains small. BlackRock puts stablecoin supply above $300 billion. Adjusted stablecoin volume passed $11 trillion in 2025. Visa's comparable volume was $11.2 trillion. Visa's other cited volume was $16.7 trillion. Stablecoins moved another $8.5 trillion in the first half of 2026. BlackRock launched a money market fund for stablecoin issuers to hold reserves earlier this year. BlackRock cites a February 2026 study by the Bitcoin Policy Institute as preliminary support for the split monetary role. The study collected 9,072 answers from 36 models from Anthropic, OpenAI, Google, xAI, and DeepSeek. The models chose stablecoins for spending 53.2% of the time. Bank money received under 9% of the responses overall. The results varied sharply by vendor. Anthropic models favored Bitcoin more often than OpenAI models. Bitcoin traded near $86,400 at press time. Bitcoin was up about 0.6% over 24 hours according to BeInCrypto Markets.