For a decade, crypto's critics had a point: the technology was too complex, too volatile, and too arcane for ordinary people to use comfortably. Every market cycle produced promises of a breakthrough consumer app, and every cycle delivered something that required a PhD to operate safely. But Alchemy CEO Nikil Viswanathan isn't defending crypto's usability record. He's making a stranger and more interesting argument: crypto was never designed for humans in the first place, and the entities it was actually designed for are finally arriving.
What Actually Happened
In April 2026, ahead of his appearance at Consensus Miami, Alchemy co-founder and CEO Nikil Viswanathan made an argument that has rapidly spread through both crypto and AI circles: the global financial system was built for humans , with operating hours, geographic boundaries, physical identity requirements, and human-readable interfaces. All of those features are bugs for AI agents. Crypto, almost by accident, was built without them. That makes blockchain the native financial infrastructure for autonomous software operating at machine scale.
Viswanathan's case is architectural, not ideological. Banks have operating hours because humans sleep. International payments face friction because humans live in specific countries with regulations tied to citizenship and physical presence. Credit cards assume a body, a government ID, and a credit history tied to a person's life. AI agents have none of these properties. They operate continuously, have no geography, and authenticate through cryptographic keys rather than names and documents. Blockchain rails are borderless, continuous, and fully programmable , and those features, which made crypto seem impossibly complicated to consumers, are precisely what makes it native infrastructure for autonomous software agents transacting at machine speed.
Why This Matters More Than People Think
The timing of this argument is significant because it arrived at the exact moment that AI agent payment infrastructure moved from theory into production. Amazon Web Services launched Amazon Bedrock AgentCore Payments in partnership with Coinbase and Stripe, enabling AI agents to purchase web content, access APIs, and transact with other agents using USDC settling on Base and Solana. Stripe-backed Tempo blockchain , co-funded by Paradigm , launched its mainnet in March 2026 with a Machine Payments Protocol specifically designed to let software agents pay for services autonomously, without requiring human approval at each transaction. These are not research projects; they are production infrastructure from companies with combined valuations exceeding $200 billion.
The developer signal is equally clear. At the EasyA Hackathon at Consensus Miami 2026, nearly 1,000 developers built AI-native startups over a single weekend, with contributors from Base and Solana ecosystems joined by engineers from Microsoft and Google. The dominant theme was autonomous payments , applications where AI agents independently initiate, authorize, and settle transactions on-chain without human intervention. Coinbase sponsored challenges specifically around x402, an emerging framework for AI agent payment authorization. EasyA founders described the event as having evolved from a crypto coding competition into a launchpad for billion-dollar companies , a framing that reflects where venture attention and developer energy are actually concentrating in 2026.
The Competitive Landscape
Multiple blockchain networks are now competing to become the default settlement layer for AI agent transactions. Solana has meaningful advantages in transaction throughput and sub-cent fees , critical for AI agents that may initiate thousands of microtransactions per hour across distributed services. Base, Coinbase's Ethereum Layer 2, offers institutional trust and regulatory clarity from its parent company's compliance posture. Ethereum is advancing ERC-8004, a new standard specifically designed for AI agent identity and payment authorization on-chain. In the stablecoin layer, USDC has emerged as the preferred denomination for agent-to-agent transactions, while PayPal's PYUSD and Tether's USDT are positioning for different segments of the emerging agent economy.
The infrastructure layer is where the most interesting competitive dynamics are playing out. Alchemy , which provides the developer tools and node infrastructure that most major crypto applications depend on , has positioned itself as the primary onramp for AI agent developers building on blockchain. Chainalysis launched blockchain intelligence agents for compliance monitoring in early 2026, capturing the institutional enforcement market. Bittensor (TAO) and The Graph (GRT) are building decentralized AI infrastructure designed to make agents less dependent on centralized providers. What makes this competitive landscape unusual is that it is not zero-sum: the total addressable market for AI agent payment infrastructure may dwarf anything crypto has previously addressed, because the number of autonomous software agents is growing faster than any prior user category in internet history.
Hidden Insight: The Real Disruption Is What Agents Do to Financial Identity
The mainstream framing of AI agents and crypto focuses on payment volume and stablecoin adoption curves. Those metrics matter, but they miss the more fundamental disruption: AI agents transacting on blockchain at scale will require , and ultimately produce , a new conception of financial identity. Traditional KYC frameworks assume the customer is a human with citizenship, age, physical address, and credit history. An AI agent has none of these. Yet it may be managing significant financial flows on behalf of its human principal, or autonomously on behalf of a protocol or organization. The regulatory frameworks for this don't exist yet, and the solutions that emerge will reshape how identity and accountability work in finance far beyond crypto.
Alchemy's on-chain credit score initiative , which went live in February 2026 , is an early signal of how this resolves. By analyzing transaction history, smart contract interactions, and counterparty patterns entirely on-chain, Alchemy built a credit scoring system for wallets rather than people. An AI agent that has successfully executed thousands of transactions without defaults or exploits accumulates a reputation score that lenders and service providers can evaluate programmatically, without any human identity verification. This is not just a crypto innovation , it is a prototype for a financial identity system that does not require governments, banks, or physical verification at all. The implications for financial inclusion are profound; the implications for financial surveillance are equally profound and opposite.
The deepest counterintuitive conclusion is about who crypto was always meant to serve. In retrospect, the "killer app" narrative that haunted every crypto cycle may have always been searching for the wrong kind of user. Crypto doesn't need a killer consumer application; it needs killer infrastructure for non-human actors. The entities that transact 24/7, across borders, programmatically, at sub-cent scale, with cryptographic authentication , those are not humans using a new financial product. They are the economic agents that the internet has been producing for thirty years and that AI is now making fully autonomous. If Viswanathan is right, crypto didn't fail to find its market. Its market simply needed AI to arrive before it could show up.
What to Watch Next
Track on-chain transaction volume attributable to AI agents specifically , not total stablecoin volume, which conflates human and agent activity. Nansen has projected that AI agents will represent the default use case for crypto by 2028, with 500 million agent wallets active on-chain. That projection implies agent transaction share should be climbing measurably through 2026. If Dune Analytics or Nansen data shows AI agent-attributed volume crossing 5% of total stablecoin transactions by Q3 2026, adoption is tracking ahead of projections. If agent volume stalls below 1%, the narrative is running well ahead of actual usage , and a correction in AI-agent-linked token valuations would follow.
In the regulatory dimension, watch for the first major enforcement action involving an AI agent rather than its human principal. The moment a regulator , the SEC, FinCEN, or the EU under MiCA , issues a fine or injunction targeting an autonomous agent, it will force immediate resolution of identity and accountability questions that are currently unresolved across the industry. That action is likely before the end of 2026. It will define the legal perimeter within which the AI agent economy can operate, and it will determine whether the current infrastructure buildout was built on solid ground or on regulatory quicksand. Companies building agent payment rails today are making implicit bets about how this resolves , those bets are worth making explicit before regulators make them for you.
Crypto didn't fail to find its consumer. It was waiting for a consumer that doesn't sleep, doesn't carry a wallet, and authenticates with a private key.
Key Takeaways
- Alchemy CEO Nikil Viswanathan argues crypto's "complexity" is a feature for AI agents , borderless, continuous, programmable rails match exactly how autonomous software operates, turning crypto's biggest consumer weakness into its core enterprise strength
- Nearly 1,000 developers built AI agent payment apps at EasyA Consensus Miami 2026 , with Coinbase's x402 framework emerging as the dominant standard and contributors from Microsoft, Google, Base, and Solana
- Amazon, Stripe, and Coinbase are in production with AI agent payment infrastructure , Bedrock AgentCore Payments settles USDC on Base and Solana; Tempo's Machine Payments Protocol launched mainnet in March 2026
- Alchemy launched on-chain credit scores for AI agent wallets in February 2026 , a prototype financial identity system for non-human actors based entirely on transaction history, with no personal data required
- Nansen projects 500 million AI agent wallets on-chain by 2028 , if that trajectory holds, AI agents will become the dominant users of blockchain infrastructure within 24 months
Questions Worth Asking
- If AI agents become the primary users of financial infrastructure, who is ultimately accountable when an agent defrauds, makes an error, or is exploited , the developer, the human principal, or the protocol itself?
- When on-chain credit scores for AI wallets become standard, does that give well-capitalized AI systems structural advantages over smaller agents , effectively creating a financial class system for software?
- If the financial system increasingly serves AI agents rather than humans, what does that mean for your ability to understand , let alone influence , where capital actually flows in the economy?