Interview: Aptos CEO on AI Agents, Stablecoins, and Machine Commerce

The rapid rise of AI agents is creating new demand for the blockchain industry: can autonomous software become an active participant in the economy, rather than just a tool for generating insights?

Franklin Templeton Digital Assets estimates that agent-based trading could reach $3-5 trillion by 2030, but significant questions remain regarding payments, security, liability, and trust.

Avery Ching, co-founder and CEO of Aptos, argues that the infrastructure supporting AI agents will need to handle fast, low-cost, programmable, and verifiable transactions.

In this interview with Invezz, Ching outlines what that infrastructure might look like and what role blockchain can play in the emerging agent economy.

Invezz: Franklin Templeton estimates that agent-based trading could reach $3-5 trillion by 2030. What needs to happen in the coming years for this prediction to become a reality?

I think the most important question is whether agents will become reliable enough for people and businesses to feel comfortable granting them authority.

The models are powerful enough for most of our daily workflows. The remaining issues are around infrastructure.

Can an agent keep your data private, can it be easily deployed, can it operate reliably for long periods, can you verify what it has done? And can it hold and exchange value according to clear rules?

For trading specifically, agents need their own identity and payment capabilities, but they also need limits.

You should be able to grant an agent access to capital with a spending limit, define who it can transact with and what it can purchase, and have a verifiable record of what happened and why.

Once you have that, the agent can move from recommending a transaction to actually completing it, and at that point, economic dynamics also come into play.

If agents are making a huge number of micro-payments, those transactions need to be fast and cheap enough that the payment itself doesn’t become the bottleneck.

Invezz: Aptos is at the top of Franklin Templeton's throughput rankings, but is TPS really the deciding factor for AI commerce, or are latency, reliability, and costs more important?

Transaction throughput is one aspect, but you have to consider the properties of the entire system.

What matters is whether the entire decentralized financial system is secure, fast, scalable, and low-cost. Aptos is the best technology to meet this high standard.

Agents are an extension of our personal and business lives.

The infrastructure needs to resemble much more mature cloud and database infrastructures: persistent state, recovery, failover, and predictable performance.

Aptos was designed specifically for a high-throughput, low-latency environment, and that’s why I believe it’s ideal for machine-driven activities.

Invezz: What is the first real use case where you foresee AI agents making payments autonomously at scale?

I think it starts with agents transacting with money.

An agent that pays for goods and services using the same interfaces we use today is the first step. Today, agents are mostly read-only in an economic sense.

They can find information and recommend an action, but usually another system or person has to complete the transaction.

Once an agent can directly pay another service on a per-request basis, a different kind of network is created.

An agent can discover what it needs, pay for it, and continue its task without opening an account, negotiating a contract, or waiting for someone to approve each individual payment.

As this grows, agents will start negotiating and transacting with each other at machine speed. That’s where I think machine-to-machine commerce will begin to become real.

Invezz: If an AI agent makes a wrong purchase or gets compromised, who should ultimately be responsible --- the user, the developer, or the underlying infrastructure?

There probably isn’t a one-size-fits-all answer, but the system should make it possible to know exactly what happened.

This is the first requirement. If an agent spends money, accesses data, or sends information somewhere, you should be able to trace the action back to the policy and authority under which it was operating, with an immutable record.

The user or business should define what the agent is allowed to do. The developer must correctly enforce those controls.

And the infrastructure should provide a record that none of the parties can silently rewrite later.

On the payment side, this means that an agent should not have unlimited access to a wallet.

It can have its own keys, spending limits, approved counterparties, or approved categories of activity.

The interesting part of programmable money is that you can give an agent enough authority to be useful without granting it unlimited authority.

As agents become more autonomous, I believe auditability becomes as important as the payment itself.

If you can’t prove what an agent did and why it was authorized to do it, companies will be very reluctant to give it significant control.

Invezz: Major AI companies already rely on centralized cloud infrastructures. Why do you think they will adopt blockchain-based payments instead of simply improving existing payment systems?

I don’t think every AI interaction needs a blockchain, and I don’t think that’s the right way to frame the question.

The question is what happens when software needs to transact with other software with which it doesn’t already have a business relationship.

Existing payment systems work very well for a person or business with an account, an identity, a payment processor, and an established counterparty.

Agents generate a different pattern. They may need to discover a service, pay a fraction of a cent or a few cents, receive the result, and move on, potentially millions of times with many different counterparties.

Opening accounts, API keys, invoices, and bilateral relationships for every interaction doesn’t scale particularly well.

Open payment channels become useful because an agent can hold value and transact programmatically under a defined policy.

Stablecoins give you a digital unit of account, and protocols like x402 make payments per request possible.

The blockchain is useful here because it provides unrelated machines with a common level of settlement and accountability without requiring a single company to manage the entire network.

Invezz: Over the years, blockchain has gone through various narratives, from DeFi to NFTs and gaming, which have taken longer than expected to reach mass adoption. What makes AI agency different, and what evidence do you see today that gives you confidence?

I would still be cautious in assuming a timeline. We are very much at the beginning. What’s changing is that the demand for agents exists independently of crypto.

People are already using increasingly capable AI systems, businesses are implementing them, and the quality of models is improving very rapidly.

The adoption of AI will be exponential when we solve the challenges of privacy, scalability, audit trails, and ease of use.

The hardest problem is transforming an agent from something that a relatively technical person can experiment with into something that an average person or a business can actually trust and depend on.

It must be private, easy to use, reliable, accountable, and capable of paying for things.

These elements already exist in some form today, but they have not yet been truly assembled into a single system.

This is what gives me confidence in the opportunity, but it is also why I wouldn’t say the market is already here.

We are still building the infrastructure that allows agents to move from demo and co-pilot to software that can operate continuously and take meaningful actions on someone’s behalf.

Invezz: This vision largely relies on stablecoins for machine-to-machine payments. How resilient is that model if a stablecoin experiences volatility or faces regulatory disruptions, and what safeguards should exist to ensure that transactions can continue safely?

I wouldn’t design the system around a single stablecoin nor assume that a single asset will work everywhere.

The important thing is that the agent has a clear policy on what it can hold, what it can spend, and under what conditions.

You can imagine a wallet for agents with approved assets, spending limits, approved counterparties, and rules that automatically halt activity if an asset or counterparty falls outside the policy.

This becomes particularly important because the goal is not to have every transaction approved by a human.

Safeguards must exist in the system before the agent acts.

Stablecoins are useful because they give software access to digital money that can move globally and programmatically, but they are only one piece of the architecture.

The most important principle is that agents have a secure way to hold and exchange value without granting them unlimited financial authority.

If we do this right, the underlying payment layer can evolve over time.