BlackRock frames stablecoins as the first payment rail for AI agents
Its paper also pitches tokenized compute as a longer-dated market that still lacks standardized, liquid contracts.
BlackRock is pitching stablecoins as the most actionable near-term crypto rail for “agentic commerce,” where autonomous software agents pay for services without a human in the loop. In the same paper, it frames tokenized claims on computing capacity as a much larger long-run opportunity that still lacks the standardization and liquidity needed to trade at scale.
BlackRock’s agentic-commerce thesis puts stablecoins in the first wave
BlackRock’s paper argues artificial intelligence could be “one of the biggest drivers for digital asset adoption” as autonomous agents start buying services, moving money, and sourcing computing power. The framing is mechanical: AI supplies “machine-native intelligence,” while digital assets supply payment and settlement rails that can execute on an agent’s decision without waiting for a person to approve each step.
In that setup, stablecoins are positioned as the first major beneficiary. BlackRock’s logic is that a stablecoin’s relatively steady price makes it usable for pricing services, and blockchains can settle payments around the clock. That combination is what makes small, automated machine-to-machine payments plausible, at least in theory, because the agent can pay for a discrete unit of work at any hour without introducing FX-style volatility into the invoice.
The paper also points to a concrete implementation path instead of leaving “AI + crypto payments” as a vibe. BlackRock highlights Coinbase’s x402 protocol as an emerging mechanism for agents to pay for online resources, explicitly including API calls. It also notes that existing payments networks are adapting to agentic commerce, which is a reminder that stablecoins are competing on settlement and programmability, not on being the only way to do machine payments.
Tokenized compute is the bigger TAM story, but market structure isn’t there yet
Where the paper gets more ambitious is compute. BlackRock describes tokenized claims on computing capacity as a longer-term opportunity, where standardized claims could eventually be traded, financed, or used as collateral through digital asset infrastructure. The bet is that if compute becomes a clean, contractible commodity, then “compute receipts” start to look like something markets can price, lend against, and warehouse.
To justify why that could matter, BlackRock cites analyst estimates that revenue from the major cloud businesses of Amazon, Microsoft, and Google could reach about $1.1 trillion by 2030. The excerpt does not identify the analysts or methodology behind that estimate, but the intent is clear: compute is a large and growing spend category, so even a small slice that becomes financeable could be meaningful.
The catch is that the paper itself flags the missing pieces. BlackRock says agent payments remain at an early stage, and that liquid markets for standardized compute contracts have yet to develop. That is not a minor footnote. Without common units, common terms, and comparable service-level obligations, “tokenized compute” risks staying a set of bespoke claims that trade like narratives rather than like commodities.
What I’m watching: x402 traction and signs of standardized compute contracts
The near-term tell is whether the rails BlackRock named start to show real usage. If Coinbase’s x402 becomes a visible pattern for paid API calls and other online resources, that is the first place you would expect repeatable onchain payment flows to show up, because the unit economics are naturally granular.
On the compute side, the threshold that matters is standardization: common contract terms, common units, and explicit SLAs that make one claim comparable to another across venues. Without that, liquidity cannot form in a way that supports financing or collateral use, which is the paper’s stated endgame.
Updates from BlackRock or other large institutions that add timelines or implementation detail would also change the read. Right now, the paper’s own characterization is “early stage,” which is accurate but not tradeable without adoption metrics.
What BlackRock: AI agents to use stablecoins Tells Me
I read this as BlackRock separating the “can settle today” rail from the “can be financialized later” asset. Stablecoins fit agentic commerce because they solve two boring problems at once: pricing in a stable unit and settling 24/7, which is exactly what machine-to-machine payments need when the transaction sizes get small.
The compute angle is bigger, but it is also where most of these stories break. If standardized compute contracts and repeatable pricing do not emerge, tokenized compute stays a collection of non-fungible promises instead of a market you can hedge, finance, or use as collateral, and the practical impact remains concentrated in stablecoin payment flows rather than a new onchain compute curve.
