In its report "The Machine-Native Economy," BlackRock suggests that the large-scale application of artificial intelligence may become an underestimated structural source of demand for digital assets, with stablecoins, native crypto assets, and other on-chain assets facilitating inter-machine settlements.

The report, authored by Will Su, Robert Mitchnick, Jay Jacobs, and William Helm, defines artificial intelligence as machine-native intelligence and digital assets as machine-native currency. The core scenario is agent-based AI: systems can plan multi-step tasks with limited human intervention, call external tools, and complete payments autonomously. The authors argue that traditional payments rely on account opening, vouchers, authorization, business hours, and human confirmation, making high-frequency trading below 1 cent uneconomical due to merchant fees, and settlement timeliness unstable.

Stablecoins are positioned as the preferred transaction layer. The report states that by September 2026, the circulating market value of stablecoins will exceed $300 billion; adjusted transaction volume will exceed $11 trillion in 2025, roughly in the range of Visa and Mastercard's annual processing volume, but still below that year's ACH of $93 trillion; from 2020 to 2025, the compound annual growth rate of adjusted transaction volume for stablecoins is about 80%, compared to about 8.5% for ACH. Price stability provides a unit of account, and on-chain near real-time settlement is suitable for interface calls, data by volume, and computational power by usage.

Payment protocols are identified as practical interfaces. The report mentions Coinbase's x402, which uses the HTTP 402 status code to allow software to make direct payments, with early use cases including USDC; Stripe's machine payment protocol with Tempo can settle interfaces and HTTP resources; Stripe's agent commerce protocol with OpenAI enables programmatic checkout while allowing merchants to retain their original cash register systems. Google and Visa also have their own agent authorization protocols. An example given is: after a user sets a $2,500 travel budget, the main agent assigns tasks to a travel sub-agent, who uses x402 to purchase ticket price data, and the main agent completes the booking through the agent commerce protocol.

The second line is the assetization of computing power. BlackRock cites analysis predicting that Amazon, Microsoft, and Google Cloud business revenues will total about $1.1 trillion by 2030, with AI capital expenditures potentially reaching about $5 trillion from 2025 to 2030. Standardized computing power rights can be tokenized for transfer, trading, financing, and collateral, with agents automatically procuring based on price, performance, and latency. The authors clarify that standardized computing power claims could become an important use case for digital assets in financing and programmable settlement, but standardized contracts and liquidity markets have yet to take shape.

In terms of market mechanisms, this is a narrative-driven configuration warm-up, not an immediate buy-sell order. Buyers include asset management institutions, stablecoin issuers, and public chain settlement layers that need to write AI infrastructure as investable themes; seller pressure falls on card organizations and bank transfer channels that rely on merchant fees and human authorization. Capital flows first target existing large-scale USD stablecoins and programmable settlement protocols, then move to non-standardized computing tokens. Beneficiaries are stablecoins and agent payment protocols already on public chains, while those under pressure are non-scaled bank tokenized deposits and traditional clearing that still charges based on manual account logic.

Supplementary data shows that model tests cited in the report indicate that in 9,072 responses from 36 leading models, about 79.1% chose Bitcoin for savings scenarios, while about 53.2% chose stablecoins for spending scenarios, with bank currency accounting for less than 9%. The actual payment volume through agents remains small, and the report describes the current stage as early.

Source: Public Information