More than $11 trillion of adjusted stablecoin transaction volume passed through the market in 2025, according to figures cited by BlackRock—an amount roughly comparable with Visa and Mastercard’s annual payment volumes. Yet that existing scale does not show that artificial-intelligence agents have already become a meaningful source of demand. It is the base from which BlackRock is making a more speculative claim: software agents may eventually need money that can move as automatically as they do.
In its “The Machine-Native Economy” paper, BlackRock Digital Assets Research argues that agents could autonomously buy data, software services and computing capacity. It presents stablecoins as the likely near-term instrument for those transactions because they can be programmed to settle around the clock between machines.
The important distinction is between a credible payments mechanism and an established new payments market. BlackRock is not saying that today’s stablecoin volumes are mainly agent-driven, nor does its report create immediate buying power for stablecoins. It is setting out a model in which recurring, low-value purchases by autonomous software become easier to execute through programmable digital cash than through workflows built around a human customer.
The $11 trillion stablecoin base is not yet evidence of agent commerce
Stablecoins already have a substantial transactional footprint. Alongside the 2025 volume figure, BlackRock’s paper cited by Decrypt points to a circulating stablecoin market capitalisation above $300 billion by September 2026. Those numbers explain why a large asset manager can plausibly treat stablecoins as payment infrastructure rather than simply as crypto-market collateral.
But the figures aggregate activity with very different purposes. They do not isolate autonomous purchases of compute, application programming interface access or datasets. The jump from broad stablecoin usage to agent-led commerce is therefore an argument about a possible change in payment behaviour, not evidence that the change has already happened at scale.
BlackRock itself acknowledges that agent payments are still at an early stage and that liquid, standardised markets for compute contracts have not developed, as CoinDesk reported. That admission matters because the thesis relies on two developments that should not be conflated: agents must make enough independent purchases to need a specialised payment layer, and the products they purchase must become sufficiently standardised to trade or finance beyond one-off service provision.
The first may emerge without the second. An agent could pay a provider for a small amount of computing capacity without creating anything resembling a financial market for computing claims. Conversely, a market for tokenised capacity would require common terms, credible settlement and participants willing to hold or finance those claims. Stablecoin settlement can support the former without resolving the latter.
x402 illustrates the payment problem agents need stablecoins to solve
BlackRock identifies Coinbase’s x402 as an emerging mechanism through which agents can pay for online resources such as API calls without a human-operated checkout. CoinDesk cites the protocol as an example of small, automated stablecoin payments.
BlackRock’s broader framing calls AI “machine-native intelligence” and crypto “machine-native money.” According to The Bid, its research argues that agents may use blockchain-based monetary instruments instead of conventional bank accounts and rails such as ACH, Fedwire or SWIFT.
The practical appeal lies in transactions that may be too small, frequent or time-sensitive for conventional checkout. An agent purchasing data, software, an API call or computing capacity could operate continuously, without making a consumer card the centre of the process. Account setup, human approval and conventional banking timetables are the constraints this model is intended to address.
The evidence does not establish a mature commercial system. Agent payments remain at an early stage, and adoption, pricing, trust and spending controls remain hurdles for providers and operators.
Stablecoins may complement bank and card rails, not replace them
The BlackRock Investment Institute’s thematic update presents agentic commerce as a combination of blockchain settlement, existing applications, payment details and third-party services—not as a clean break with the current payments system. Different rails can therefore serve different parts of the same transaction landscape.
Stablecoins may be useful when an autonomous agent needs a native, programmable way to pay another online service. Many transactions, however, may remain inside established applications or rely on payment credentials and intermediaries already embedded in merchant systems. The practical question is whether stablecoins offer enough operational advantage in a particular agent workflow to justify changing the surrounding stack.
That framework does not imply that agent adoption would necessarily erode card networks or bank rails. Stablecoins could add machine-to-machine settlement while substantial activity remains on existing infrastructure, complementing transactions that are awkward under human-centred payment design rather than universally replacing established payment systems.
Tokenized compute is the larger financial-market bet—and the least developed link
The potentially larger idea in BlackRock’s framework is not a stablecoin payment for an API call. It is tokenised compute. The firm says standardised claims on computing capacity could eventually be traded, financed or used as collateral, extending the argument from payments infrastructure to a prospective market in an input that AI systems need.
That prospect sits against an enormous possible revenue pool. BlackRock references estimates that Amazon, Microsoft and Google cloud revenue could reach about $1.1 trillion by 2030, according to Ledger Insights. The number gives the thesis its financial ambition: if compute becomes a standardised, transferable claim, an expanding cloud market could support more than straightforward service payments.
Yet scale of cloud revenue is not evidence that capacity can readily be packaged into a liquid instrument. The missing market structure is central, not incidental. Compute services vary by provider, configuration, location, timing and contractual conditions. Turning access to those services into claims that buyers can confidently trade, lenders can finance and markets can accept as collateral requires far more standardisation than a stablecoin transfer itself.
BlackRock’s acknowledgement that such liquid standardised markets do not yet exist puts tokenised compute in a different category from the near-term stablecoin use case. The former is a proposed financial layer atop cloud infrastructure; the latter can begin with a seller, an agent and an agreed payment method. The two ideas reinforce each other conceptually, but their development paths are unlikely to be equally fast.
BlackRock has supplied institutional validation, not new stablecoin buying power
The immediate significance of the report is narrative and institutional, rather than transactional. It is not a new BlackRock product or capital commitment, and therefore does not directly create incremental stablecoin demand. InvestingLive characterised the near-term effect as validation of the AI-agent and digital-asset narrative rather than a direct market catalyst.
That validation is still notable. A firm of BlackRock’s scale is framing stablecoins not only as crypto-market instruments but as a candidate payments layer for a growing class of automated software. But the investable-looking part of the story remains furthest away: a tokenised compute market that can be standardised, financed and used as collateral. For now, the more concrete signal is that the payment experiment has begun before the financial market built around it exists.
Disclaimer: This article is provided for informational purposes only. It is not offered or intended to be used as legal, tax, investment, financial, or other advice.
Be the first to comment