BlackRock report links AI agents to a possible $5 trillion stablecoin opportunity

BlackRock says AI agents could become a new kind of stablecoin user

BlackRock says artificial intelligence could create a new type of customer for stablecoins: software that pays for things on its own, without a person approving every payment. A stablecoin is a digital token built to hold a steady value, usually linked to the US dollar.

The asset manager makes the case in a new report called “The Machine-Native Economy.” The report says increasingly independent AI systems could buy data, access software and rent computing power by themselves, adding a source of payment demand for digital assets beyond trading and ordinary human payments.

CryptoSlate reported on the report on September 23, 2026.

Key numbers from the report

• BlackRock says autonomous AI systems could make thousands of very small payments, around the clock, to finish a single task.
• Stablecoins had more than $300 billion in circulation and about $11.2 trillion in adjusted transaction volume in 2025, according to the report.
• The report estimates that volume grew at an 80% compound annual rate between 2020 and 2025, compared with roughly 8.5% for the US Automated Clearing House network.

How stablecoins compare with older payment rails

The ACH network, a US system for moving money between banks, still processed about $93 trillion last year, according to the report. That figure shows how far stablecoins remain from the largest traditional payment systems.

BlackRock also warned against comparing stablecoin activity directly with Visa and Mastercard, because those networks count transactions in different ways.

Payment firms are building competing standards

Several companies are already working on ways for software to pay for services. Coinbase’s x402 protocol uses the web’s HTTP 402 “Payment Required” status code to ask for payment before returning data or another resource. An AI agent can request an API — a way for one program to use another program’s service — receive payment instructions, send USDC, and get the service without a person completing a checkout.

Stripe and Tempo are developing the Machine Payments Protocol, which can settle payments in stablecoins or through traditional methods. Stripe and OpenAI’s Agentic Commerce Protocol links AI agents to existing merchant systems, while Google and Visa are working on separate standards for agent identity and authorization.

Because the approaches differ, the report does not assume machine commerce will automatically move onto blockchains. Traditional networks can adapt when agents deal with established businesses and consumers, while stablecoins may fit better where payments are extremely small, very frequent, or built directly into software.

Where the money could end up

BlackRock says a second question is where the value from those payments lands. If agents create more stablecoin transactions on Ethereum, that could raise demand for blockspace and for validator services. ETH is used in the network’s fee and staking system, which is one route by which higher activity can affect the native asset.

But the report says transaction growth and demand for a token do not necessarily rise together. It says the amount captured by native crypto assets will depend on fee structures, staking economics and gas-sponsorship models, where an app covers the network fee for the user. Networks can handle large volumes while charging very little, and applications can keep users and agents from holding the gas token themselves.

Circle’s Arc, a blockchain focused on payments, works differently. It uses USDC as its native gas asset, so more activity there could strengthen the stablecoin’s role without creating the same link to a separate token such as ETH.

Computing power as the next market

BlackRock expects the same payment setup to reach one of AI’s biggest costs: computing power. The report says cumulative investment in AI infrastructure could exceed $5 trillion between 2025 and 2030. It also cites Bloomberg consensus forecasts that put combined revenue from Amazon Web Services, Microsoft’s Intelligent Cloud business and Google Cloud at about $1.1 trillion by 2030.

In that scenario, an agent could compare computing providers by price, hardware, location, latency or performance, buy capacity for a specific task, and settle the cost automatically. Payments could happen per job, per use, or per model token, creating a repeating machine-to-machine loop of finding, buying, using and paying for computing power.

BlackRock also raises the idea of a larger financial market forming around that activity. Standardized claims on computing capacity could eventually be traded or pledged as collateral, and futures markets could let buyers and sellers hedge changes in compute costs. The report says such markets would need standards that account for differences between chips, energy prices, locations and performance.

What is confirmed

BlackRock published the report “The Machine-Native Economy,” and the figures and market descriptions above come from the report as covered by CryptoSlate. The stablecoin volume, growth rate, ACH comparison, AI infrastructure investment estimate and cloud revenue forecast are all presented as BlackRock estimates or as forecasts the report cites.

What is still unclear

The report itself describes the compute market idea as largely prospective. Agentic payment activity is still in its early stage, and traditional financial companies are building their own systems for autonomous commerce alongside crypto firms.

It also remains unsettled whether machine payments will grow enough to matter for token prices or network economics, and which networks or standards will handle the largest share of that traffic. BlackRock does not state that higher activity will automatically raise the value of any specific token.

Sources

• CryptoSlate