Invest1 publisher2 min readPublished
BlackRock hands institutional investors the crypto industry's AI demand argument
In The Machine-Native Economy, four BlackRock researchers argue agentic commerce will favour stablecoins and that claims on computing capacity could be tokenized and pledged. Coinbase's Brian Armstrong made the case in July.
The Investor · Invest desk
What happened
- BlackRock, the world's largest asset manager, says in new research that broad AI adoption could be an underappreciated source of demand for digital assets.
- The paper is titled The Machine-Native Economy and is credited to BlackRock's Will Su, Robert Mitchnick, Jay Jacobs and William Helm.
- It argues that existing payment rails can support some automation while account setup, credentialing and authorization could still require human involvement.
- It also says merchant fees can make low-value transactions uneconomic and that settlement and finality times could vary from provider to provider.
- On compute, it says AI companies could seek to lock in costs and providers, and that claims on that capacity could be tokenized, transferred, pledged as collateral or traded.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- precedent Cointelegraph notes the crypto industry has argued this link for years; a BlackRock research note makes it quotable in an institutional memo, so the argument now has to survive allocator diligence rather than conference applause.
- decision Naming stablecoins as the likely leader in transactional use points the allocation question at issuers and payment rails, and away from broad token baskets.
- constraint A tokenized claim on compute can be transferred the day it is minted, but it cannot be pledged until a lender will value GPU hours and enforce delivery on a default.
- exposure Once crypto infrastructure is filed as AI-adjacent, it competes for the same allocation as chip and data-centre exposure, and gets judged on agent settlement volumes nobody currently publishes.
The demand case rests on a friction list and one instrument choice. The authors name stablecoins, native cryptocurrencies and tokenized real-world assets as suited to high-frequency, sub-cent machine-to-machine transactions running around the clock [8]. "Together, these developments position AI as a structural catalyst for digital asset adoption and digital assets as a potential facilitator of the AI economy," they wrote, adding that the relationship "remains underappreciated and could expand the role of digital assets as core infrastructure for an increasingly autonomous digital economy" [4][5].
Take sub-cent literally. A million payments at under a cent each settle less than 10,000 dollars in total [19]. The count can multiply while the settled value stays trivial, and the paper is specific about which instrument it expects to carry the traffic: "Several types of digital assets may support agentic commerce, but stablecoins are likely to lead transactional use," the authors said [9].
The supply side shipped before the research note. Cointelegraph's account names four products built for this work: Coinbase's x402 protocol, Tempo's Machine Payments Protocol, Circle's agent wallets and USDC payment tools from May, and OKX's Agent Payments Protocol, which handles recurring payments and escrow released when a task completes [16][17][18][20]. Brian Armstrong, Coinbase's chief executive, made the demand argument in July, pushing back on calls for crypto to pivot to AI [14]. "AI being a megatrend takes nothing away from crypto," he wrote, and "If anything, it makes crypto more important" [15].
The compute leg is the harder half. The authors expect AI agents to buy resources automatically in these markets, and say tokenized capacity "could in turn broaden institutional investor participation and establish compute as a new opportunity for the broader digital asset ecosystem" [12][11]. A traded claim needs someone quoting both sides of it. Cointelegraph's report does not include a size for the compute market or a volume figure for agent payments.
In my view the transactional half of this is checkable within a year and the compute half is not. The counter-thesis is that the friction BlackRock lists is a pricing choice incumbents can reverse: cut per-transaction economics for machine traffic, automate credentialing, and the sub-cent argument thins out. Confirmation would be dull and specific, published settled value on x402, Tempo and OKX's protocol, denominated in stablecoins and growing quarter on quarter. Falsification is a compute forward market that clears in dollars on bilateral contracts with no token anywhere in it.
What to watch
- Published settled value on x402, Tempo's Machine Payments Protocol or OKX's Agent Payments Protocol, which would make the transactional claim measurable.
- A tokenized claim on computing capacity that trades at a quoted price with an enforceable delivery obligation behind it.
- Whether BlackRock follows the research note with a product that takes balance-sheet exposure to either leg.