Alchemy CEO Nikil Viswanathan is making a provocative but increasingly credible case: the global financial system was designed around human behavior, and as AI agents become the primary economic actors on the internet, crypto is the only financial infrastructure built to accommodate them.
Most arguments for crypto adoption focus on humans, the unbanked, the privacy-conscious, the inflation-hedging investor. Nikil Viswanathan, who runs Alchemy, one of the most widely used blockchain infrastructure platforms in the industry, is pointing in a different direction entirely. The next wave of economic activity, he argues, will not be driven by humans opening wallets. It will be driven by AI agents executing transactions autonomously, at machine speed, across borders, around the clock. And the traditional financial system, built on bank accounts, identity verification, business hours, and human-legible processes, is structurally unsuited to support that.
The argument is worth taking seriously because it comes from someone with direct visibility into how developers are actually building on blockchain infrastructure right now. Alchemy processes billions of dollars in transaction volume and serves a developer base that spans DeFi protocols, NFT platforms, and increasingly, agent-based applications. When Viswanathan says he is watching the use cases shift, that observation carries operational weight that a theoretical argument would not.
Consider what an AI agent actually needs to transact. It needs to open accounts without a government ID. It needs to send value programmatically without a human approving each transfer. It needs to operate across jurisdictions without navigating a patchwork of correspondent banking relationships. It needs to settle instantly, because a multi-day clearing window is not a minor inconvenience for a system executing thousands of micro-transactions per hour. It needs to do all of this without a legal entity, a credit history, or a relationship manager at a financial institution.
The traditional banking system fails on every single one of those requirements. Not because of regulatory malice or technical incompetence, but because it was designed around the assumption that the entity transacting is a human being or a human-controlled organization. That assumption is now breaking down at precisely the moment when agentic AI systems are moving from research environments into production workflows with real economic consequences.
Crypto, by contrast, has no account opening process. A wallet is generated algorithmically. Transactions are authorized by private keys, not identity documents. Settlement on most major networks now happens in seconds, not days. Smart contracts execute payment logic automatically when predefined conditions are met, with no human required to approve the transfer. These properties were often framed as advantages for privacy-seeking humans or DeFi speculators. Viswanathan's point is that they are actually more naturally suited to machine-to-machine commerce than to anything humans typically do.
What Agent-Native Commerce Actually Looks Like
The practical applications are closer than the framing might suggest. AI agents already browse the web, execute code, manage calendars, and interact with APIs on behalf of users. The next logical extension is agents that pay for the services they consume, negotiate contracts with other agents, distribute revenue across contributor networks, or manage treasury functions for decentralized organizations without a human CFO in the loop. All of those workflows require programmable, permissionless financial infrastructure. Crypto provides it. Stripe does not.
Several projects are already building explicitly in this direction. Protocols designed for agent-to-agent payments, on-chain reputation systems that allow agents to establish creditworthiness without human identity, and smart contract frameworks that let agents enter binding economic agreements are all in active development. The ecosystem is early, but the architectural logic is sound. Developers who want to build AI agents that can transact autonomously are reaching for crypto rails because there is no comparable alternative in traditional finance.
The regulatory dimension adds complexity. Financial regulators have not yet addressed the question of how existing frameworks apply when the transacting entity is an AI agent rather than a legal person. That ambiguity creates real friction for enterprise adoption, but it does not change the underlying technical reality that Viswanathan is describing. If anything, the regulatory gap is likely to resolve in ways that formalize crypto's role in agent commerce rather than restrict it, because the alternative is asking AI systems to route through a banking infrastructure that simply cannot support their operating requirements.
For investors watching both the AI and crypto sectors, the convergence thesis Viswanathan is articulating is worth building into their models now rather than waiting for it to become consensus. The infrastructure layer that wins the agent commerce market, whether that is Ethereum, Solana, or a purpose-built network, will capture a volume of transactions that makes current DeFi activity look modest. Alchemy, sitting at the developer infrastructure layer, is positioned to be a direct beneficiary of whichever chain wins that race. That context is worth keeping in mind when evaluating the argument being made.
Also read: Solana Foundation is lending USDT into Aave and bringing the AAVE token to Solana as DeFi's biggest lending protocol fights for its footing • Printr is turning its omnichain launchpad into a public company by letting anyone buy in for $200 • Luck.io bet $500,000 a month on influencer marketing, processed $1.2 billion in wagers, and shut down in eleven months
