According to BlackRock, artificial intelligence agents require “machine-native money,” with stablecoins standing out as the top candidate. A recent research publication from the world’s leading asset manager contends that traditional bank transfers and card networks were never designed for software interacting financially with other software.
The study outlines a dual-layer monetary framework for machines. Stablecoins manage everyday spending, whereas Bitcoin functions as the designated savings vehicle, based on research referenced by BlackRock.
Because software transactions for computing power, data, and services cannot afford to wait a full business day for standard bank clearance, BlackRock proposes a currency capable of settling at machine speeds, placing stablecoins at the forefront.
Why BlackRock Says Cards and ACH Fall Short for AI Agents
The foundation of BlackRock’s perspective lies in a logistical challenge. An autonomous program cannot independently establish a card or bank account without human intervention. Additionally, standard merchant processing fees render micro-payments under one dollar impractical, while Automated Clearing House (ACH) transactions typically require up to a business day for final settlement.
Conversely, blockchain networks operate continuously around the clock, achieving settlement nearly instantaneously. Consequently, the report concludes that on-chain assets represent a natural match for automated machine transactions.
“stablecoins, native cryptoassets, and other on-chain assets can serve as machine-native instruments” BlackRock said
The firm highlights Coinbase’s x402 protocol, which revitalizes the long-dormant HTTP 402 “Payment Required” response code to let software agents pay for information immediately. BeInCrypto has previously covered this x402 standard along with the subsequent surge in token activity.
Nonetheless, BlackRock does not view this exclusively as a cryptocurrency monopoly. The publication acknowledges competing financial rails developed by Visa, Google, OpenAI, and Stripe—some of which rely on traditional fiat currency—while also noting that current transaction volumes for active software agents remain modest.
Even so, the sheer magnitude of the stablecoin market reinforces the thesis. BlackRock estimates the total circulating supply exceeds $300 billion. Adjusted transaction volume reached roughly $11 trillion in 2025, closely matching Visa’s $11.2 trillion, though remaining trailing Visa’s $16.7 trillion figure.
During the opening half of 2026, stablecoins processed an additional $8.5 trillion.
BlackRock is already active within this ecosystem, having introduced a specialized money market fund earlier this year designed to hold reserves for stablecoin issuers.
AI Models Pick Bitcoin to Save and Stablecoins to Spend
The document’s most widely discussed conclusion is borrowed from outside research. BlackRock references a February 2026 investigation conducted by the Bitcoin Policy Institute (BPI), an advocacy and research organization focused on bitcoin, treating it as “preliminary support” for a divided monetary structure.
The BPI evaluated 36 frontier artificial intelligence models provided by OpenAI, Anthropic, Google, DeepSeek, and xAI, gathering a total of 9,072 individual responses. When asked to select a store of value, the models chose Bitcoin 79.1% of the time. When prompted on what currency to use for expenditures, stablecoins were selected in 53.2% of instances, whereas traditional bank money accounted for less than 9% overall.
BeInCrypto previously covered the release of this AI Bitcoin preference analysis. Two important distinctions should be kept in mind: the outcomes varied significantly across different developers, with Anthropic’s models displaying a notably stronger preference for Bitcoin than those created by OpenAI.
At the time of writing, Bitcoin (BTC) changed hands near $86,400, reflecting a 24-hour gain of approximately 0.6% according to BeInCrypto Markets data.
Readers should also note that predictive statements within the report consistently utilize conditional phrasing like “can” or “could,” alongside an explicit disclaimer noting that the text does not constitute a formal forecast or financial advice.