BlackRock, the world’s largest asset management company, has released new research stating that the rapid adoption of artificial intelligence (AI) could become a driving force behind digital asset demand that the market has not yet given adequate attention. As AI agents and machines evolve toward machine-to-machine payments, stablecoins, blockchains, and other on-chain assets may become important payment infrastructure supporting a new digital economy.

AI agents drive demand for "machine-native" payments

BlackRock, in its latest research report (The Machine-Native Economy), points out that AI and digital assets may form a two-way relationship: advances in AI will increase demand for blockchain and programmable payment infrastructure, while digital assets may provide payment, trading, and resource-allocation tools for the AI economy. The research team believes this connection has not yet received sufficient attention, and in the future it may further expand the role of digital assets as infrastructure in a highly autonomous digital economy.

Baird believes that the widespread adoption of AI agents may change traditional payment models. Existing financial payment systems can support a certain degree of automation, but processes such as account setup, identity verification, and payment authorization often still require human involvement; in addition, transaction fees may make large volumes of low-value transactions economically inefficient, and settlement speed and finality can differ across payment service providers.

In contrast, stablecoins, native cryptocurrencies, and tokenized real-world assets (RWAs) have characteristics such as operating 24/7 and being programmable; in theory, they are more suitable for AI agents to conduct high-frequency, sub-cent machine-to-machine trading. Baird expects that although many digital assets may support AI-agent economies, stablecoins are most likely to become the primary trading tool first.

This also means that in the future, AI agents may not just help humans complete tasks; they may also be able to pay for API, cloud services, data, software, and other digital resource costs themselves, further forming economic activities that are executed autonomously by machines.

AI compute power could become a new crypto asset market

Besides payments, Baird is also focused on the core resources required for AI development—computing power. As demand for AI training and inference grows rapidly, companies may want to lock in future GPU and other compute resource prices in advance, while compute providers need to manage risks arising from demand and price fluctuations.

Baird proposes a possible market model: tokenizing the right to use future computing power or the right to its revenue, so that related rights can be transferred, traded, and even used as collateral. In this way, compute power would not only be AI infrastructure, but could gradually become a financialized asset that can enter the digital-asset market for trading.

Baird believes that once this kind of market takes shape, it may not only help AI companies manage compute costs more efficiently, but also attract more institutional investors to participate, further developing “compute power” into a new asset class within the digital-asset ecosystem. At the same time, AI agents themselves may directly participate in these markets, automatically purchasing compute resources based on actual demand.

Crypto industry players have begun laying the groundwork for AI-agent payments

Baird’s view also aligns with the direction of development in the crypto industry in recent years. Coinbase CEO Brian Armstrong said this past July that AI will not diminish the importance of cryptocurrencies; instead, it may increase demand for “programmable money,” because AI agents need financial tools that can execute transactions on their own.

So far, multiple crypto companies have begun building related infrastructure. Coinbase has launched the x402 protocol, and Tempo is developing the Machine Payments Protocol—both targeting the demand for AI agents to automatically pay for online services. Circle has also introduced wallets and USDC payment tools that support AI agents, while OKX has proposed the Agent Payments Protocol, supporting applications such as recurring payments and escrowed payments.

Overall, Baird’s research extends the relationship between AI and digital assets beyond a mere investment theme, further into payment, compute, and resource-trading infrastructure. If AI agents can eventually execute transactions at scale on their own, the use cases for stablecoins and other on-chain assets may expand from currently human-led trading to digital economic activities in which machines directly participate.

Source