BlackRock has laid out a case that the next major source of demand for stablecoins and blockchains may not be human investors at all, but software. In a September 2026 whitepaper titled The Machine-Native Economy, the firm’s Digital Assets Research group argues that artificial intelligence and digital assets are converging into two halves of the same system: AI as “machine-native intelligence,” and crypto assets as “machine-native money.”
The paper is authored by Will Su, BlackRock’s Head of Digital Assets Research, alongside Head of Digital Assets Robert Mitchnick, U.S. Head of Equity ETFs Jay Jacobs, and Head of U.S. iShares Product Innovation William Helm. Its central claim is that as AI systems become “agentic,” able to carry out multistep tasks with limited human oversight, they will need a way to pay for things natively, and that traditional payment infrastructure is poorly suited to the job.
The Core Argument: Two Kinds Of Tokenization
BlackRock’s framing rests on an analogy between how large language models and blockchains each work. Both, the paper says, rely on “tokenization” that translates “information and economic entitlements into discrete, standardized representations that machines can process natively.” An LLM converts human language into numerical tokens it can compute over; a blockchain represents value and ownership as digital-asset tokens that machines can verify and transfer without an intermediary.
That symmetry, the authors argue, is why the two technologies fit together. An AI agent that can reason but cannot transact is only half a participant in the economy. Connect it to a blockchain, and it can both decide and pay.
Why Not Existing Payment Rails
The paper contends that legacy systems — ACH transfers and card networks — carry friction that makes them awkward for autonomous, high-frequency machine commerce. It cites “account setup, credentialing, and authorization processes that may require human involvement,” merchant fees that make “very low-value transactions uneconomic,” and settlement delays unsuited to “always-on, very low-value transactions requiring programmable execution.”
Stablecoins and cryptoassets, by contrast, settle around the clock, can carry programmable conditions, and can support very small payments. The paper points to a cluster of emerging protocols built for exactly this: x402, Coinbase’s protocol that repurposes the dormant HTTP 402 “Payment Required” status for blockchain-agnostic settlement; ACP (the Agentic Commerce Protocol from Stripe and OpenAI); MPP (a Machine Payments Protocol from Stripe and Tempo); Google’s A2A and AP2 agent-coordination standards; Anthropic’s Model Context Protocol (MCP); and Visa’s Trusted Agents Protocol. BlackRock also reads Stripe’s August 2026 acquisition of OpenRouter as a signal that “model routing and compute-usage optimization are becoming part of the financial infrastructure surrounding AI.”
Some of that machinery is no longer hypothetical. As The Crypto Times has explained in its primer on x402, the protocol is now stewarded by an x402 Foundation co-founded by Coinbase and Cloudflare, with support from Google, Visa, AWS and Anthropic, and it recorded a roughly 100,000% jump in transaction volume in a single month around late October 2025; a spike off a very low base rather than a sustained run-rate, but an early sign of developer interest. The broader industry has moved in the same direction: Coinbase has pushed AI agents into crypto trading and payments, and Binance has begun pitching AI agents as a new class of exchange customer. BlackRock’s paper is, in that sense, an institutional articulation of a thesis crypto-native firms have been building toward for the past year.
The Numbers BlackRock Leans On
To size the opportunity, the paper marshals several data points. It puts the stablecoin market capitalization above $300 billion as of September 2026, and cites roughly $11 trillion in adjusted stablecoin transaction volume in 2025, a figure it says now sits “in the same broad range as Visa and Mastercard’s annual payment volumes” (around $27 trillion combined) while remaining “well below the $93 trillion” moved over ACH. It notes stablecoin volumes grew at roughly an 80% compound annual rate between 2020 and 2025, against about 8.5% for ACH.
The paper also frames compute itself as “a distinct, large, and increasingly investable economic resource.” It cites projections that hyperscaler cloud revenues could exceed $1 trillion annually by 2030, with combined revenue from AWS, Microsoft’s Intelligent Cloud and Google Cloud estimated near $1.1 trillion by then, a 29% annual growth rate from 2025, and that AI inference will become the largest AI workload by 2030. In BlackRock’s telling, agents could eventually “autonomously source and pay for compute over blockchains” using tokenized claims on capacity.
These are projections and framings, not established outcomes, and BlackRock presents them as such.
The Honest Caveat Inside The Paper
Notably, BlackRock does not claim the machine-native economy has arrived. The demand is prospective. The paper’s illustrative workflows — an agent booking travel across calendars and sub-agents, or sourcing compute from a marketplace and settling per use — are explicitly hypothetical.
It also flags the thinness of the evidence on how agents would actually choose to hold money. Citing recent Bitcoin Policy Institute research, the paper notes that “model outputs across controlled simulations generally favored stablecoins for everyday payments and bitcoin for long-term value preservation,” pointing to “a potential AI-native monetary architecture,” but immediately qualifies that these are “simulated model responses rather than observed agent behavior.” In plain terms: the rails have been built faster than the agents have shown up to use them, and the little evidence there is comes from models role-playing in a lab, not autonomous software transacting at scale in the wild.
On regulation, the paper situates its thesis against a friendlier backdrop, citing the U.S. GENIUS Act, the stablecoin law whose first year The Crypto Times examined in this analysis of how it is reshaping U.S. frameworks, along with the EU’s MiCA framework and stablecoin regimes in Hong Kong and Singapore.
Reading The Source: BlackRock’s Own Stake
One piece of context matters for how this whitepaper should be weighed. BlackRock is not a neutral observer of digital-asset adoption: it is the world’s largest asset manager and a direct commercial participant in the market it is describing. It runs the iShares spot bitcoin and ether exchange-traded funds, launched BUIDL, its first tokenized fund for institutions, on public blockchains in 2024, and has since signaled ambitions to tokenize its ETFs and cash-management ranges. A thesis that AI agents will generate structural new demand for stablecoins, blockchains and tokenized assets is, among other things, a thesis that supports the firm’s own product direction.
That does not make the analysis wrong; the parallels the paper draws are real, and the protocol activity it points to is verifiable, but it is a reason to read the projections as an interested party’s outlook rather than disinterested fact. BlackRock says as much in its own disclaimers. The document states it is “not intended to be relied upon as a forecast, research or investment advice” and “not a recommendation, offer or solicitation to buy or sell any securities,” adding: “There is no guarantee that any of these views will come to pass.”
Why It Matters
The significance of the paper is less any single number than the source. When the largest asset manager in the world publishes a formal thesis that autonomous software could become a structural buyer of stablecoins and blockspace, it moves the “AI-meets-crypto” narrative from crypto-native circles toward the institutional mainstream, and gives allocators a framework for a demand story that, until now, has lived mostly in developer forums and token pitches.
Whether that demand materializes, and on which chains, in which stablecoins, under which rules, remains unproven. For now, as BlackRock’s own hedging makes plain, the agents have been given the rails faster than they have shown up to use them. The Crypto Times makes no forecast on if or when that changes.
Also Read: ARK and Securitize Tokenize Venture Fund With OpenAI and Anthropic Exposure
