Crypto Built for AI Agents, Says Alchemy CEO

Nikil Viswanathan, CEO of blockchain infrastructure provider Alchemy, argued on Apr 25, 2026 that the architecture of public blockchains is intrinsically better suited to autonomous crypto-payments" title="Coinbase Sees AI Agents Driving Crypto Payments">AI agents than to human-first financial workflows (Coindesk, Apr 25, 2026). His comments position crypto primitives—programmable money, composable smart contracts, and on-chain identity—as the native rails for machine-to-machine commerce, not merely human-mediated transactions. Viswanathan's thesis challenges long-standing assumptions about user experience as the primary determinant of blockchain adoption, and instead reframes the debate around machine-native semantics, transaction composability and verifiable state. The proposition has immediate implications for infrastructure providers, cloud vendors, exchanges and custody solutions as agents could compress settlement, authentication, and counterparty discovery into automated, always-on flows. For institutional investors, the shift suggested by Alchemy's CEO demands a re-evaluation of which technology stacks and service providers will capture value as agent-led activity scales.

Context

Viswanathan’s remarks build on a multi-year evolution in which developer activity and protocol design have progressively prioritized composability and machine-verifiable state over human-centered interfaces. Alchemy itself became one of the most prominent infrastructure firms after raising $200 million in January 2022 at a $10.2 billion valuation (TechCrunch, Jan 2022), underscoring the market’s earlier confidence in infrastructure monopolies for Web3 development. The company’s position as a middleware provider gives it an informed vantage point on developer behaviour: historically, developers gravitate to platforms that reduce friction for automation and integrate easily with external compute and data sources. That developer dynamic is the foundation for Viswanathan's argument that AI agents—capable of autonomously negotiating, settling and composing on-chain transactions—represent the next phase of commerce.

The claim must be set against the backdrop of broader AI adoption forecasts. PwC estimated in 2017 that AI could contribute up to $15.7 trillion to the global economy by 2030, a number frequently cited in strategic planning discussions for technology and finance firms (PwC report, 2017). Whether that value accrues to cloud compute, AI model providers, or to transaction rails like blockchains depends on interoperability, latency, and the cost of atomic settlement. In short, a large macroeconomic tailwind for AI does not automatically translate into on-chain activity without technical and economic incentives that favour verifiable, decentralized settlement.

A practical comparison helps illustrate the point: markets evolved from human dealers to electronic order books and then to algorithmic and high-frequency trading over the last three decades. Algorithmic trading now accounts for a majority of volume in many developed equity markets (historical U.S. equity data, 2010s), which establishes a precedent where machine-native workflows dominated once latency, liquidity and regulation matured. If AI agents follow a comparable trajectory, the underlying rails will matter more than human-facing UX for a subset of transactions that can be fully automated.

Data Deep Dive

Viswanathan’s assertion rests on observable trends in developer tooling, infrastructure demand and transaction design. Alchemy’s Jan 2022 capital raise ($200m at $10.2bn) is one tangible data point that reflects investor conviction in infrastructure capture (TechCrunch, Jan 2022). More recent usage metrics from middleware providers show rising API call volumes and node requests year-over-year—signals that programmatic, rather than purely human, traffic is increasing on public ledgers (company reports, 2024–2026). These telemetry patterns are consistent with an ecosystem where bots and backend services, not only wallets used by end-users, generate a significant share of on-chain events.

On the compute side, AI model costs and hardware availability are relevant constraints. The cost of running advanced generative models has declined materially since 2023 due to model optimizations and more competitive GPU supply, which lowers the marginal cost for an agent to perform off-chain intelligence before executing an on-chain action. That margin compression creates an economic case for agents to perform complex decisioning off-chain and then leverage on-chain primitives for settlement, accountability and composability. This hybrid architecture—off-chain inference and on-chain settlement—maps to current enterprise designs where cloud providers and blockchain nodes coexist.

Comparative metrics matter for investors assessing where value will accrue. Public cloud providers (MSFT, GOOGL, AWS) capture recurring revenue from AI workloads, whereas blockchain middleware captures API and node traffic for transaction submission and state queries. Exchanges and custody platforms (e.g., COIN) capture flows associated with fiat on-ramps and large-value settlements. If agent-led commerce grows 10–30% annually in transaction volumes over the next five years, middleware and custody could see disproportionate revenue growth relative to decentralized protocol token markets, which are more volatile and less directly tied to transaction-processing revenue. Those growth bands are illustrative but grounded in current developer telemetry and cloud usage trends (industry reports, 2024–2026).

Sector Implications

Infrastructure: Middleware and node operators stand to benefit from agent-native traffic because agents will require predictable, low-latency RPCs and richer query primitives. Firms with diversified revenue models (API fees, enterprise SLAs, data products) will be better positioned than token-only models to monetize a steady increase in machine-originated calls. Investors should watch companies that report usage-based billing and enterprise contracts rather than purely token metrics.

Cloud & AI stack: Major cloud vendors that offer both AI model hosting and blockchain node services could capture a high-margin slice of the stack. Microsoft, Google, and Amazon already cross-sell cloud compute for blockchain analytics and indexers; an agent economy would deepen that integration as agents demand co-located compute and transaction finality. Hardware suppliers—principally NVIDIA (NVDA)—remain critical because faster inference at lower cost makes agent decisioning more feasible for mass-market use cases.

Exchanges & custodians: Exchanges and trusted custody providers could see new flows as agents require reliable fiat rails, custody solutions, and programmable settlements. The net effect may be to increase centralized activity at the rails layer even as certain application-level functions remain decentralized. Coinbase (COIN) and other regulated entities could benefit from custody and brokerage demand, though outcomes will depend on regulatory clarity and enterprise adoption.

Risk Assessment

Regulation: Agent-native commerce raises regulatory questions that differ from current debates centered on retail consumer protection and token classification. Liability attribution—who is responsible when an autonomous agent executes a harmful or fraudulent transaction—remains undefined in most jurisdictions. Regulatory responses could slow adoption if frameworks require onerous KYC/AML processes for every autonomous counterparty or if liability regimes force custodians to act as gatekeepers.

Security and MEV: Machine actors may accelerate extractive behaviours like MEV (miner/extractor value) unless protocol-level mitigations evolve. The interaction between agents competing in the same mempool could amplify front-running risks and lead to more sophisticated contestation strategies, increasing on-chain gas price volatility. Vulnerabilities in smart contracts that were benign in human-mediated scenarios could be weaponized by agents executing at scale.

Concentration risk: A paradox of agent-native commerce is that its economics can favour consolidation. Agents will seek reliable counterparties, low-latency nodes, and predictable settlement—the very attributes that large incumbents provide. That can create winner-take-most dynamics for middleware and custody, counter to crypto's decentralizing narrative. For market participants, concentration increases counterpart risk and regulatory exposure.

Fazen Markets Perspective

We view Viswanathan’s thesis as directionally credible but structurally complex. The technical primitives of blockchain—atomic settlement, programmable trust, and public verifiability—are indeed well-suited to machine-to-machine contracts. However, the economic capture will be bifurcated: recurring enterprise revenue for middleware and custody providers is a nearer-term, higher-confidence outcome than token appreciation across volatile markets. This implies that listed equities tied to cloud, AI hardware, and regulated custody (e.g., NVDA, MSFT, COIN) could be more direct plays on agent adoption than protocol tokens.

Contrarian insight: Agent-driven commerce could paradoxically accelerate centralization rather than decentralization. Agents optimize for reliability and latency; as a result, a handful of middleware and cloud providers could mediate most agent activity, recreating Web2-like chokepoints inside a Web3 facade. That consolidation would shift regulatory scrutiny to these intermediaries, and could create cross-market linkages that make crypto markets more correlated with cloud and AI equities than they are today.

Practical investment implication: Monitor usage-based revenue metrics from middleware providers, enterprise SLAs signed by custody firms, and API-request growth as leading indicators of agent adoption. Publicly reported developer telemetry and enterprise contract disclosures will likely be higher-fidelity signals than spot token prices.

Outlook

Time horizon: We expect meaningful agent-native on-chain activity to materialize in discrete pockets over the next 3–7 years rather than explode systemically in months. Early use cases will likely be constrained to low-friction, high-frequency applications—dynamic hedging, automated market-making strategies, indexed NFT management—where agents can demonstrate clear economic advantage. Broader retail-facing agent adoption will require regulatory guardrails and user trust, which tend to evolve more slowly.

Adoption scenarios: In an optimistic scenario—where regulatory clarity emerges and infrastructure scales—middleware revenues and custody flows could grow at double-digit compound annual rates through 2028. In a conservative outcome, regulatory friction and security incidents slow enterprise adoption, pushing most agent activity into private, permissioned environments that do not materially benefit public blockchain economics.

Signals to watch: For investors tracking this thesis, key indicators include quarterly developer activity reports from middleware firms, enterprise SLA announcements, CPU/GPU pricing trends for inference workloads, and regulatory signals on liability for autonomous agents. We also recommend tracking mempool dynamics and MEV auction data as early technical signals of agent behaviour in public networks.

Bottom Line

Alchemy’s CEO frames a plausible narrative: blockchains provide primitives that are naturally compatible with autonomous agents. The more likely near-term winners are infrastructure and custody providers that monetize machine-originated activity, while token market outcomes will depend on adoption breadth and regulatory response.

Disclaimer: This article is for informational purposes only and does not constitute investment advice.

FAQ

Q: How quickly did machines displace humans in a comparable finance context, and what does that imply for agents on-chain?

A: Electronic and algorithmic trading transitioned over decades; algorithmic strategies accounted for a majority of U.S. equity volume by the early 2010s, after the proliferation of electronic order books and colocation. The implication is that agent-led adoption on-chain is likely to be uneven and multi-year—markets need time to mature, latency must improve, and regulation must catch up before agents dominate.

Q: Will agent-native commerce necessarily increase token prices?

A: Not necessarily. Increased on-chain activity can boost fee capture for nodes and middleware, which benefits service revenue streams more directly than token markets. Token appreciation depends on tokenomics—supply elasticity, burn mechanisms, and market expectations—so higher transaction volumes do not automatically translate into token price increases.

Q: What operational metrics should investors monitor?

A: Track API call volumes, enterprise contract wins reported by middleware or custody firms, GPU utilization and pricing trends for inference workloads, mempool transaction composition changes (share of automated submissions), and public disclosures on MEV mitigation and protocol upgrades.

blockchain infrastructure | AI investment themes