BlackRock research points to AI agents as hidden crypto demand driver

BlackRock, the globe's biggest money manager, has put out a research note contending that artificial intelligence could be a demand source for digital assets that is being overlooked. Titled The Machine-Native Economy, the paper was produced by the firm's digital assets group, with Robert Mitchnick and Will Su among the authors. It represents a research thesis rather than an actual purchase, filing or product launch, and it stops short of asserting that such demand has materialised.

BlackRock's argument is as follows.

Within the paper, AI is characterised as machine-native intelligence, while digital assets are framed as machine-native money. The premise is that autonomous AI agents — software capable of planning and executing multi-step operations with minimal human oversight — will require spending on data, services and computing power rental. Legacy financial systems like banks and card networks were designed around human users, accounts and click-based confirmations. BlackRock contends that stablecoins and other on-chain instruments are well-suited for the small, round-the-clock payments these agents would make. The firm lists stablecoins, native cryptocurrencies and tokenized real-world assets as potential vehicles, and also puts forward tokenized computing capacity as an additional possibility.

Why the paper matters, and what it does not claim.

Payments, not price speculation, form the mechanism at work. Stablecoins offer the most straightforward match, since an agent purchasing a few seconds of compute needs a stable unit to measure cost. Reporting on the paper highlights that it does not contend Bitcoin will become the standard currency for machine-to-machine transactions, so the document should not be read as a bullish call on Bitcoin.

How much of this is in place today.

Evidence suggests the volume is minimal. TRM Labs, a blockchain analytics firm, looked at $52.7 million in settlements on Coinbase's x402 protocol — a system that allows software to pay for an API call in the same request — and calculated that AI agents probably accounted for 0.6% to 7.5% of that total. BlackRock itself concedes that agent payments remain in their infancy.

Two cautionary points come from critics. A TechFlow analysis asserts that the narrative is hardly novel in the crypto space, and that on-chain data markets have generated barely any notable revenue for the data-purchasing behaviour the paper emphasises. The analysis also notes that agent payment protocols are in competition with one another. For instance, Stripe and Tempo's Machine Payments Protocol settles in both stablecoins and fiat currency, meaning traditional payment networks could retain a portion of the traffic.

What to watch for going forward.

The immediate measurable checkpoint is whether observed agent activity climbs past that 0.6% to 7.5% range on x402 and comparable protocols. A rise in absolute agent-driven volume would lend support to the thesis. Should the agent share remain in the low single digits, the paper stays a projection rather than a demand narrative. If protocols that settle in both fiat and stablecoins capture the bulk of agent traffic, the argument for crypto-specific demand would be undermined.

For readers, the useful interpretation is to view the paper as a guide to which infrastructure warrants monitoring, rather than as an indication of price movement. A BlackRock research note does not constitute proof of institutional purchases, and no transaction is being suggested.