Julien Maldonato (Deloitte): "Stablecoins are money with legs that never sleeps"

Julien Maldonato (Deloitte): "Stablecoins are money with legs that never sleeps"
Ask AI TO SUMMARIZE ThIS ARTICLE

Julien Maldonato, partner at Deloitte and specialist in financial infrastructure, explains why the convergence of agentic AI, stablecoins and blockchain is reshaping the rails of the global financial system.

Your 2 free articles this month are up

The research your peers are already leveraging

The Big Whale gives financial institutions the market intelligence, network, and platform to move with confidence in digital assets. Trusted by 150+ firms.

You have been working in the infrastructure sector for nearly thirty years. You seem to harbour a certain frustration with the pace at which technology standards penetrate the world of finance. How do you view this evolution?

My frustration is real and it goes back to 1994. At the time, I was discovering the HTTPS protocol and I could already see the 402 error code, which was supposed to be dedicated to digital payments. Yet it took thirty years for the XML format to finally be adopted by Swift through the ISO 20022 standard.

We are dealing with timescales that seem disconnected from the reality of innovation. Decisions are not always rational or scientific; they are often political or emotional. It is a bit like the nuclear energy debate. That said, there are reasons for optimism: we are finally seeing an awakening at the European level. The hostility towards new infrastructure that could be felt ten years ago is fading. People are beginning to understand that the real issue is the infrastructure, not just the asset.

You often refer to a "triple convergence" when defining blockchain. Can you explain this concept for a traditional finance audience?

It is a pedagogical effort I try to lead. In my view, blockchain is the unification in a single point of three pillars: the Ledger, the Network and the Rules. Historically, these functions are separated. Swift manages the network, commercial and central banks maintain the ledgers (the accounts), and the rules are dictated by law and enforced by trusted third parties such as lawyers.

In a public blockchain, these three functions converge within a single line of code. The network becomes the ledger and it is self-executing. It is a contraction of space-time that is still difficult for many to grasp, because it challenges historical silos. If you fail to understand this, you risk creating a fragmented, proprietary "Internet of money", akin to the Minitel or the Chinese model.

Beyond this infrastructure, you see agentic AI as the "final variable" in this equation. Why now?

Because everything is becoming granular. In 1994, Jeff Bezos created Amazon in his garage. His great insight was selling books individually anywhere in the world. Today, thirty years later, we are about to relive this phenomenon with nano use cases.

Institutional finance has become "retailised", and we are now heading towards the "retail of retail". In the 1970s, we saw markets become electronic and high-frequency trading (HFT) emerge. We are about to experience the equivalent of HFT in everyday consumption. Autonomous agents will navigate across different exchanges on our behalf, possibly using units other than the euro. It is a reinvention of barter, made fluid by AI, which enables the "double coincidence of wants" to be found instantaneously.

Why do these AI agents have a vital need to handle money autonomously?

Exchange is the common thread of our societies. To explore new territories or structure new trades, you need to make information flow more freely. Money is nothing more than an information system that makes value and prices legible.

Today, there are too many opaque areas. The AI agent brings a capacity for direct action where, previously, human validation was required. We will delegate signatures to our wallets, just as we learned to delegate payments to our contactless cards. Without exchange, an economy dies. The autonomisation of transactions, even for micro-amounts, is the indispensable shift from conversation (the chatbot) to transaction (the agent).

What are the concrete benefits for the real economy?

We will move beyond the era of crude monthly subscriptions that do not reflect actual usage. The agent will enable ultra-fine adjustment to physical reality. Take the example of inventory management: an agent can monitor your fridge or a factory line and trigger the exact transaction at the right moment.

Our current systems, steered by central banks that move rates in massive increments, are too imprecise. Agentic AI makes it possible to create local virtual markets, as close to the need as possible. We are deconstructing the large infrastructures and the big SaaS subscriptions to move towards total granularity.

You argue that blockchain is the "natural rail" for this autonomous AI. Are current payment infrastructures not sufficient?

The current Internet has become hyper-centralised. It is performant, but fragile. Above all, it is not a "ledger". To pay on the Internet, you still need to go through about a dozen intermediaries (central banks, commercial banks, processors, etc.), which creates friction, delays and costs.

The next stage of computing is hyper-connectivity that becomes a global ledger. It is the "great world computer" that Vitalik Buterin described ten years ago. Two building blocks were missing: selective confidentiality (for compliance) and the interface. Agents solve the interface problem. Handling a wallet or tokens may seem technical for a human; for an agent, it is its native language. Blockchain then becomes a physical inevitability.

Is blockchain the only infrastructure capable of effectively serving these AI agents, or could other rails emerge?

This is a complex question because we still lack hindsight. What is certain is that the collapse of transaction costs on new networks, where "gas" is virtually free, is a compelling argument. Being able to execute a transaction for a millionth of a dollar is an unbeatable value proposition for an agent.

I would not be binary enough to say it is the only path, because a technological breakthrough is always possible. But today, it is the main one. We are witnessing a deconstruction of the old world to save time and money. The US administration seems to have understood this: it is willing to challenge giants like Visa or Mastercard. If American public and private powers are going "all in" on blockchain, it is because the candidate is virtually unique at this stage.

Can you give us concrete use cases that would resonate with financial decision-makers?

Historically, stablecoins have been used for cross-border transfers. But tomorrow, the real issue will be the programmable settlement asset. Imagine a payment that automatically sets aside an insurance premium or loyalty points.

Take insurance or escrow: AI can manage millions of micro-insurance contracts across an entire subcontracting chain, from a Chinese factory to your doorstep. Today, insuring every link in that chain is a monumental task. With agentic AI, you can audit and trace these flows in real time. The agent becomes an auditor capable of processing quantities of information that are humanly impossible to handle.

You also mention the identification of objects. What is the link with digital assets?

This is a major topic: the Digital Product Passport (DPP). A European regulation is coming within two to four years. In practice, if you buy a bicycle, you will have full traceability of parts, maintenance and repairability.

These proofs will be encrypted. The passport will take the form of a wallet, as the company Arianee had anticipated several years ago. In the same environment, you will have your settlement proofs, your identity proofs and your proofs of objects. It is this "cryptographic recognition" that will accelerate exchanges. We are not just gaining in volume, we are gaining in speed (frequency). It is the "Red Queen" theory: if you do not run faster, you die.

Can we imagine investment funds entirely managed by agents, without human intervention?

Trading algorithms have been doing this since the late 1990s. But agentic AI goes further. It will put an end to money "sleeping" in current accounts.

As soon as your settlement assets become mobile and autonomous (via stablecoins), there is no reason for them to remain idle. They will finance an invoice, a loan or an investment on the other side of the world, around the clock. Stablecoins are money with legs that never sleeps. This opens the door to impact investing: we will finally be able to prove, through data captured by agents, that the investment is genuinely effective.

Who are the players leading the charge today? Circle often comes to mind...

Circle has a very modern vision and has managed to convince giants like Amazon to join the adventure (via AWS). But there is also Stripe, which is well ahead. Shopify could also emerge as a winner.

Unfortunately, Europe is lagging somewhat. Our stablecoin issuers are often start-ups of thirty people. In France, we have gems like Morpho (DeFi) or Zama (confidentiality), but large corporates are still hesitant to shift their cash management to stablecoins. Why? Because the treasurer of a major group does not yet have a structured traditional banking offer that proposes stablecoins. We are waiting for the corporate and investment banks (CIBs) to bridge the gap.

Swift has announced it wants to become a blockchain hub...

That is very good news. Swift has European roots, and this is something to watch closely. We are going to experience a long period of cohabitation between the old rails and the new ones. In retail, players like Walmart or Etsy are already experimenting with agentic commerce. Tomorrow, the agent will be your advisor: it will not just find the product, it will negotiate and trigger the payment.

You mentioned Amazon. Concretely, how does the agentic AI plus stablecoins combination fit into their business model? Is it an opportunity or a threat for such a platform?

For a giant like Amazon, it is a double-edged sword. On one side, there is a disintermediation threat. Amazon has already had friction with certain AI agents (such as Claude) that were "scraping" their site to make purchases on behalf of the user.

The risk for Amazon is that the customer experience becomes purely transactional and "cold". Their entire strategy rests on suggestion, cross-selling and visual appeal. If an AI agent simply compares prices technically and places an order via an API, the entire marketing lever collapses. That is why they are seeking to lock in the experience through their own agents, like Alexa, and to control the entry points.

And on the strictly financial side? We know that merchants have been waging a historic war against banking fees.

This is precisely where the stablecoin becomes the ultimate weapon. It is the perpetual fight against interchange commissions and acquisition fees, which can eat up to 2% or 3% of revenue.

Today, if Amazon can acquire a financial flow directly via a stablecoin like Circle's USDC (or even through a future "Amazon Coin") without going through traditional card payment rails, it means savings of several billion dollars. They have already tried with Amazon Pay, but it was a closed-loop system that struggled to gain traction. Stablecoins offer the advantage of being programmable and integrating into an open ecosystem. Ultimately, Amazon does not just want to be a merchant, but a financial and logistics advisor capable of managing your flows in real time.

We have talked a lot about infrastructure, but one protocol is making a comeback: the X402, recently revived by Coinbase. Why is this standard, which technically dates back to the early days of the web, suddenly becoming the keystone of the agent economy?

It is a masterstroke in terms of infrastructure strategy. The HTTP 402 error code ("Payment Required") was designed from the very origins of the Internet, but it had remained an "empty shell" because we did not have a currency native to the network. Coinbase had the audacity to dust off this standard and turn it into the common language between a service emitter and a service receiver.

In practice, X402 allows an agent to access an API or a web service without needing to create an account, subscribe or go through manual KYC. The agent sends a request, receives a "402" response, signs a micro-payment in stablecoins (typically USDC on Base or Solana), and instantly accesses the resource. It is the end of the rigid SaaS model for machines. By being at the origin of the standard, Coinbase is positioning itself to become the "Google of payments": they do not own everything, but they define the rules of the game.

Google is not sitting idle either with its AP2 standard (Agent Payments Protocol). Is this head-on competition or complementarity?

It is complementary. X402 handles the pure transaction: "I owe you $0.001, here is the signature." Google's AP2 tackles what comes before: authorisation and trust. Through the W3C's Verifiable Credentials (VCs), AP2 makes it possible to create cryptographic "mandates".

This is essential for risk management: how do you prove that an agent actually received the order from its owner to buy a specific plane ticket? The mandate is an inviolable digital contract that accompanies the request. Google and Coinbase are actually working together so that X402 becomes the payment extension of AP2. We are witnessing a massive standardisation where identity (Google) and settlement (Coinbase/Circle) are merging.

In this race for speed, Ethereum sometimes seems to lag behind Solana, which is posting transaction records. For "autonomous finance", is the choice of rail already settled?

We need to be careful with binary thinking. Solana has an obvious advantage in speed and cost, which makes it an ideal playground for "everyday HFT" and micro-transactions. But Ethereum remains the security and deep liquidity layer of DeFi.

The future will probably play out on Layer 2s (like Base) that try to reconcile Ethereum's robustness with Solana's agility. What is certain is that interoperability will be the final arbiter. You cannot rely on a single rail. We are going to experience a somewhat complex "multi-chain standard" period, but a necessary one before reaching unification, much like USB-C eventually prevailed for hardware.

To conclude, AI is often discussed as a risk, but you seem to view it as a driver of economic equity.

AI is stochastic; it can generate errors or chaos. Cryptography is its antidote: it provides determinism and verification. It is the antidote to deepfakes and fraud.

Sam Altman and others often speak of the "society of abundance". The idea is that AI produces value at near-zero cost, and that digital assets make it possible to redistribute this value in a granular and equitable manner. The stablecoin is not merely a speculative tool; it is the infrastructure that enables the remuneration of millions of micro-contributions, whether human or robotic. It is a total paradigm shift for finance: moving from the management of massive, slow flows to the management of an infinity of instantaneous micro-flows.

Grégory Raymond

Grégory Raymond is Head of Research and co-founder of The Big Whale. A specialist at the intersection of traditional finance and digital assets, he has been covering the regulatory, institutional and technological developments of the sector since 2017 for an audience of decision-makers: ,banks, asset managers and fintechs. He is also the author of Bitcoin & Cryptos: L'enjeu du siècle (Talent Éditions, 2025), a book built around interviews with key figures from the ecosystem.

See all articles ↗
People in the article
Topics
Business
Download our latest benchmark
Benchmark 2026 : AI Agents x Stablecoins: the payment infrastructure for the autonomous economy
Download
Subscribe to The Drop
The leading weekly briefing on digital assets for financial institutions: independent analysis, reports, benchmarks and exclusive events, delivered to your inbox.
Read by 30,000 professionals

Ready to accelerate your digital asset strategy?

Contact Us  →