Crypto Long & Short: What will the AI agents run on?

Utkarsh Ahuja makes the case that an AI-driven economy — in which autonomous agents negotiate, purchase compute, pay for data and execute transactions — will require financial rails that can move value as programmatically as information currently moves. He suggests blockchain technology and digital assets are natural fits for that role because they let money be embedded in code: wallets, smart contracts and stablecoins permit machine-initiated transfers without the manual layers typical of legacy payments. Ahuja singles out stablecoins as an important bridge between programmable onchain value and familiar units of account, and notes tokenization is already migrating traditional assets onto blockchain rails. He argues that combining programmable intelligence with programmable assets could unlock new economic activity, since an AI agent that can analyze markets but cannot efficiently hold or settle assets is limited. Identity, provenance and custody are other infrastructure needs Ahuja highlights. As autonomous agents act on behalf of humans, markets will require ways to verify who or what is acting, what permissions they hold, and where data and intellectual property originated. While blockchains will not solve every aspect of these problems, their capacity to provide shared, verifiable records makes them a useful component of the stack. Ahuja cautions investors against indiscriminate exposure to projects that attach AI and crypto labels without solving real problems. He recommends focusing on infrastructure that addresses genuine frictions created by increased automation — payments, settlement, identity, cybersecurity, custody and the interfaces between traditional and digital markets — rather than assuming value will accrue solely at the application layer.