BlackRock published a report this week titled "The Machine-Native Economy," arguing that AI agents could become the next major source of demand for cryptocurrency. The $15 trillion asset manager says autonomous software programs, which need to pay for things like API calls and on-demand compute around the clock, are a natural fit for blockchain rails rather than traditional payment systems.
The report is notable because of who wrote it. BlackRock isn't a crypto-native shop, but it hasn't been a skeptic for a while. The firm's iShares Bitcoin Trust won SEC approval in 2024 and went on to have the most successful ETF debut on record, now managing over $67 billion in assets.
Why card networks don't work for software
The core of BlackRock's argument is plumbing. Card networks and automated clearing houses were built for humans — you sign up, you get onboarded, you pay a fee that makes sense for a $40 dinner but not for a fraction-of-a-cent API call. Settlement is also slower than an AI agent's decision loop. If a machine is buying data or compute by the second, waiting on batch settlement is a nonstarter.
BlackRock says crypto-native rails handle the opposite: high-frequency, sub-cent, machine-to-machine transfers. Think API calls, on-demand data, consumption-based compute. The report frames this as a gap traditional finance hasn't addressed because it was never designed to serve non-human customers.
Stablecoins for spending, bitcoin for saving
The report points to what it calls an AI-native monetary architecture, and it's a split you've heard before in human crypto circles. Stablecoins do the transactional work. Bitcoin stores value. BlackRock cites Bitcoin Policy Institute research finding that in controlled simulations, AI models generally favored stablecoins for everyday payments while preferring bitcoin for long-term value preservation.
That's not a claim that models have feelings about money. It's a finding about how they're trained and prompted to optimize, and BlackRock treats it as a signal about what agentic commerce might default to as these systems get more capable.
A familiar thesis in a new wrapper
BlackRock has been warm on crypto infrastructure for years. It has talked up asset tokenization repeatedly and described bitcoin as an asset class of its own, the kind investors buy to hedge against potential debt crises. The new report doesn't break from that view. It extends it, arguing that as AI adoption broadens, digital assets become increasingly integral to AI's economic infrastructure — stablecoins, tokenized real-world assets, and native cryptoassets that settle on blockchains.
Nothing in the report commits BlackRock to launching a product for AI agents. It's a research note, not a fund filing. But the firm's research arm tends to signal where its business is looking, and the ETF business already showed it will follow a thesis into a filing when the demand is there.
The open question is whether agentic payments actually scale in the near term, or whether this stays a compelling architecture diagram until AI agents have real budgets to spend. For now, the most concrete thing BlackRock has done on this front remains its bitcoin ETF — over $67 billion in assets and a debut that set the bar for everyone else.
