BlackRock: AI Agents Need Machine-Native Money—Bitcoin as Savings?

BlackRock has argued that AI agents require “machine-native money,” with stablecoins emerging as the leading option. In a new research paper, the world’s largest asset manager contends that card networks and bank transfers were never designed for software transacting with other software.

The paper outlines a two-tier monetary framework for machines. Stablecoins are used for spending, while Bitcoin serves as the savings vehicle, according to a study BlackRock references.

Software that independently purchases data, compute, and services cannot afford to wait a full business day for a bank transfer. BlackRock’s proposed solution is money that settles at machine speed, with stablecoins named as the primary choice.

Why BlackRock Sees Cards and ACH as Inadequate for AI Agents

BlackRock’s reasoning begins with a concrete issue. An autonomous agent cannot open a bank account or obtain a card without human involvement. Merchant fees make micro-payments uneconomical, and ACH transfers still require up to a business day to clear.

In contrast, blockchains settle nearly instantly and around the clock. The paper concludes that on-chain assets are therefore a natural fit for machine-to-machine transactions.

“stablecoins, native cryptoassets, and other on-chain assets can serve as machine-native instruments” BlackRock said

The firm highlights Coinbase’s x402, a protocol that resurrects the dormant HTTP 402 status code so agents can pay instantly for data. BeInCrypto has previously covered the x402 payment standard and the subsequent token market excitement.

However, BlackRock does not present this as a crypto-dominated future. The paper lists competing frameworks from Stripe, OpenAI, Google, and Visa, some of which rely on traditional bank money. It also acknowledges that live agent payment volumes remain modest.

The scale of stablecoins lends weight to the thesis. BlackRock estimates the circulating supply at over $300 billion. Adjusted volume surpassed $11 trillion in 2025, nearly matching Visa’s $11.2 trillion, though it remains behind Visa’s $16.7 trillion.

In the first half of 2026, stablecoins processed an additional $8.5 trillion.

The asset manager is already active in this space. Earlier this year, it launched a money market fund designed for stablecoin issuers to park their reserves.

AI Models Favor Bitcoin for Saving, Stablecoins for Spending

The paper’s most attention-grabbing claim is borrowed. BlackRock cites a February 2026 study by the Bitcoin Policy Institute (BPI), a pro-bitcoin research and advocacy group, as “preliminary support” for a divided monetary role.

The BPI team tested 36 frontier models from Anthropic, OpenAI, Google, xAI, and DeepSeek, gathering 9,072 responses. When asked where to store value, the models picked Bitcoin 79.1% of the time. For spending, they preferred stablecoins 53.2% of the time. Bank money attracted less than 9% of overall choices.

BeInCrypto covered the AI Bitcoin preference study when it was first released. Two caveats are noteworthy. The outcomes differed substantially by vendor, with Anthropic models favoring Bitcoin far more frequently than OpenAI’s.

Bitcoin (BTC) was trading near $86,400 at press time, up roughly 0.6% over the past 24 hours, according to BeInCrypto Markets.

It is important to note that every forward-looking statement in the report uses “could” or “can,” and a disclaimer clarifies it is not a forecast. Therefore, this should not be considered financial advice.