Pantera Capital Partner: The Financial Trajectory of AI Agents

Original Author: Cosmo Jiang

Compiled by: Ken, Chaincatcher

AI agents are becoming economic participants. However, what kind of financial rails do they need to conduct transactions without human intervention?

The viral rise of OpenClaw (formerly known as Clawdbot) marks a generational leap in autonomy. When these AI agents begin to interact with each other—sometimes even negotiating and trading—the future of agents shifts from science fiction to actionable reality.

OpenClaw is just one step in an accelerating development process. Trillions of dollars are being invested in building an AI-powered world. By 2026, spending by major cloud service providers in the AI sector in the U.S. alone is expected to exceed $650 billion, about ten times the inflation-adjusted cost of the Apollo program.

Once simple chatbots are now evolving towards agentification and fully autonomous systems. These AI agents will not only generate content but also become economic participants; they will be able to reason, act, trade, coordinate, and debate, all without real-time human supervision.

Some predictions suggest that by 2030, AI agents could facilitate $3 trillion to $5 trillion in global consumer business transactions. Even if only 10% of these transactions occur between agents, it would mean hundreds of billions of dollars in machine-native settlement flows each year.

This naturally raises the question: what kind of financial and coordination rails are suitable for the native commerce of AI agents?

Today's business systems are built for humans, involving personal identification, banking intermediaries, legal contracts, settlement cycles, and human oversight. Autonomous software cannot walk into a bank branch, sign paper documents, or wait days to complete ACH (Automated Clearing House) settlements. What agents need is an infrastructure that is inherently programmable, always online, globally accessible, permissionless, and machine-verifiable.

Blockchain can meet these constraints, and we are already seeing this dynamic emerge.

In line with the rapid rise of OpenClaw in January, the trading volume and active addresses on Solana have also begun to climb. Signs on the social network Moltbook, where the AI agents reside, suggest they may have driven this growth.

The x402 payment protocol developed by Coinbase allows agents to make real-time payments for digital resources without accounts or complex identity verification. Since its launch, usage has steadily increased.

Currently still in its early stages, existing examples are more directional than definitive. However, if investors are excited about AI innovations, it would be a mistake to overlook "why blockchain rails may become the cornerstone for unlocking fully autonomous agents."

Levels of Autonomy

Many will rightly point out that today's AI agents do not need blockchain. This is true, but it is somewhat shortsighted.

McKinsey outlines six levels of automation in AI-driven business, from basic assistance (Level 0) to fully autonomous inter-agent commerce (Level 5). Levels 0-4 can operate within existing financial rails because transactions still rely on human identities. Human users have already been verified on ChatGPT, Amazon, or Perplexity, and linked their credit cards. Agents merely act as proxies, inheriting the identity, payment credentials, and legal status of that human.

The infrastructure for this model—shared payment tokens, chargeback systems, fraud detection—already exists and operates well through Visa or Stripe.

At Level 5, blockchain rails will become crucial: at this point, agents will transact directly with other agents without human instruction; there will be no human identity to inherit; payments must be programmatic, conditional, and settled within milliseconds; and agents' reputations must be portable across different platforms.

As long as humans are involved and accountable, traditional rails are sufficient. However, once agents become economically independent participants, the constraints will change fundamentally.

Agent Finance

To understand where value will converge and why blockchain is important, we must envision the ultimate logical form of agent AI. Some agents will be created by companies or individuals. Others will be generated by agents themselves, forming increasingly independent systems capable of reasoning, capital allocation, and trading without real-time human supervision.

Without human-specified transaction channels (such as going to a bank, using Stripe, or establishing a blockchain wallet), agents will rationally choose rails that maximize speed, reliability, and global coverage while minimizing friction and dependencies. When the alternative is to open a bank account and wait for ACH settlements during limited banking hours, agents will naturally opt for permissionless, always-on blockchain rails.

We see three structural constraints that will drive agents toward blockchain rails.

Identity and Access

Before agents can transact, counterparties must know who or what they are dealing with.

Traditional identity systems are designed for humans, relying on government IDs, physical signatures, and legal personhood. Autonomous AI agents possess none of these.

Binding agents to human bank accounts immediately raises questions: How do you conduct anti-money laundering (AML) checks on software? Who is the liable entity? How do you independently authorize multiple agents? How do you isolate misconduct without freezing an entire account?

In simple cases, agents can inherit their owner's credentials (e.g., ChatGPT Checkout). But this model will collapse when scaled. Agents need their own verifiable identities rather than borrowing from humans.

Blockchain-based identity verification allows agents to prove their authorization without disclosing sensitive information. You can think of it as a digital credential that anyone, anywhere can verify instantly, without calling a lawyer or querying a database.

Emerging standards like Ethereum's ERC-8004 propose on-chain registries where agents can establish verifiable credentials and accumulate transaction history and reputation over time. An agent with thousands of undisputed transaction records is fundamentally different from a brand-new agent, and this reputation is portable across different platforms.

Trust is a prerequisite for business. In an agent-driven economy, the core question will shift from "intercepting rogue bots" to "identifying which bots are trustworthy."

Programmable Money and Micropayments

Traditional payment rails are designed for human-scale transactions. Credit card fee structures make micropayments below a cent unfeasible. Anti-fraud systems also flag high-frequency machine behavior as suspicious.

Commercial activities between agents operate on a completely different scale. An agent writing code may execute thousands of API calls within a single workflow. Another may compare prices among hundreds of data providers. Payments must occur within milliseconds, often amounting to just fractions of a cent.

On-chain transactions can be divided into extremely small units, with very low settlement costs. More importantly, they are programmable. Payments can be conditional: only pay when data is valid, release funds upon task completion, or stream payments in real-time as services are consumed. Agents can cryptographically prove their solvency without pre-funding accounts, greatly enhancing capital efficiency.

Blockchain enables financial infrastructure that aligns with how agents operate: autonomous, high-frequency, conditional, and capital-efficient.

Deterministic Execution

Traditional business builds trust on intermediaries. Banks guarantee settlements. Payment processors manage chargebacks. Courts adjudicate disputes. Contracts ultimately rely on human legal systems.

When billions of low-value transactions occur across different jurisdictions, this framework comes under immense pressure. Non-human participants may not share jurisdiction, legal remedies, or enforceable contracts. Cross-border enforcement is often slow, costly, and fraught with uncertainty.

Blockchain encodes enforcement directly into smart contracts, reducing reliance on centralized systems or legal remedies. Settlements are deterministic and not subject to subjective interpretation. Rules are transparent and verifiable in advance. This is what blockchain enthusiasts refer to as "trustless execution."

For autonomously operating agents at scale, minimizing reliance on centralized intermediaries can reduce friction and increase predictability. Lower friction also expands the boundaries of feasible economic activity. Agent commerce empowered by blockchain rails has the potential to accelerate global GDP growth.

This is Just the Beginning

The question is not whether agent commerce will arrive, but on what infrastructure it will operate.

As AI agents become autonomous economic participants, the number of participants in the global economy will grow exponentially. Agents will need digital-native financial rails, capable of handling programmatic settlements, massive micropayments, permissionless coordination, and trust-minimized identities. These principles are at the core of blockchain design.

We believe that the rapid proliferation of AI agents is a strong long-term structural tailwind driving blockchain activity. Preliminary evidence suggests this is happening, and we believe it represents a value creation opportunity that is underestimated by most investors.