BlackRock published a report called "The Machine-Native Economy," and the pitch, as reported by Bankless on September 23, is not a vague crypto-meets-AI slogan. It's narrower and more useful: AI agents are about to generate a volume and shape of payment traffic that existing rails weren't built for, and blockchains plus stablecoins are the closest fit anyone's shipped.
That's a claim worth taking seriously precisely because it's specific. Here's what BlackRock actually argues, and where the reasoning holds up.
The argument: agents pay differently than people do
Per Bankless's reporting, BlackRock's framing is that "AI agents will increasingly need payment infrastructure built for 24/7, high-frequency, often tiny transactions for APIs, data, and compute — or, in other words, blockchains and stablecoins."
Strip the branding and it's a plumbing argument. A human paying for a SaaS subscription tolerates a monthly card charge, a bank holiday, a $0.30 processing fee. An agent calling a dozen APIs per task, every few seconds, around the clock, does not tolerate any of that — the fee structure and settlement delay of legacy rails don't scale down to machine-frequency, machine-sized payments. That's a real gap, and it's the same gap x402 and similar HTTP-native payment protocols were built to close: pay-per-call, settled in stablecoins, no account setup, no batch billing cycle.
Worth being precise about what's claimed here versus shown: this is BlackRock's projection of what infrastructure agents *will need*, not a report of agents currently transacting at scale on these rails. The demand is directional, not yet measured.
The bigger bet: financializing compute itself
The more interesting part of the report is about compute, not payments. BlackRock expects markets to emerge for "pricing, financing, hedging, and trading access" to compute resources — with "standardized claims on future compute capacity that can be tokenized, transferred, used as collateral, and settled programmatically."
The justification is a supply-side number: BlackRock cites forecasts showing inference workloads rising from 25% of data-center power demand in 2025 to 43% by 2030. If that holds, compute stops being a thing you provision in advance and becomes a thing that fluctuates constantly with agent demand — training-style bulk allocation gives way to a "much more fragmented, continuous market," per the report. A continuous, fragmented market for a scarce resource is exactly the kind of thing that gets financialized: futures, collateralized claims, programmatic settlement. BlackRock is, unsurprisingly, an asset manager describing a future asset class.
This is the part to flag as forecast, not fact. Tokenized compute futures don't exist as a liquid market today. The 25%-to-43% inference share is a cited projection, not an observed outcome, and "can be tokenized" is a capability claim, not a deployed product.
Where x402 actually fits
The one concrete mechanism named in the reporting is x402: agents that "shop for compute in real time and pay for it per use, per job, or even per model token" through the protocol. That's the connective tissue between the two halves of the argument — the payment-rail case and the compute-market case meet at the same place, a protocol that lets a machine pay another machine for a resource without a human approving each transaction.
It's also the part builders can act on today, independent of whether the tokenized-compute-futures thesis ever materializes. Per-call, stablecoin-settled payment infrastructure is buildable now; a market for compute derivatives is not. If you're building an agent that needs to pay for APIs, data, or inference on the fly, the pattern BlackRock is describing already has working implementations to build against.
What to watch
Whether "machine-native economy" turns into deployed volume or stays a well-argued thesis depends on adoption we can actually measure: real agent-to-agent payment traffic over protocols like x402, and whether any compute marketplace ships a tokenized claim that trades. Until then, treat this as one of the better-reasoned institutional cases for agent payment rails — and a reminder that the stack an agent needs (wallet, payment protocol, spend controls) is infrastructure you can map and check today, independent of whether the bigger compute-market bet pays off.
Sources
- ▸[BlackRock Maps Crypto's Role in the AI Economy](https://www.bankless.com/read/blackrock-maps-crypto-s-role-in-the-ai-economy) — Bankless, September 23, 2026