About 18 months ago, I wrote about AI’s growing role in crypto. The thesis was simple: artificial intelligence would enhance crypto markets through better analytics, fraud detection, and trading strategies. AI market cap sat at just 1% of total crypto and the opportunity seemed obvious.
However, I was looking at it backwards.
The Inversion
The real story isn’t about AI tokens or AI-powered trading bots. It’s about something far more fundamental. Crypto isn’t a vertical within the AI economy. It’s the payment rails the entire AI economy will run on.
This shift became clear over the past few months as the AI agent economy started moving from theory to reality. We’re not talking about chatbots anymore. We’re talking about autonomous agents that book flights, negotiate deals, search the web and complete transactions without human intervention. These agents work 24/7/365, across every time zone, and they transact at a scale humans simply cannot match.
The question nobody was asking: how do billions of AI agents actually pay each other?
Why Traditional Finance Breaks
Here’s where it gets interesting. The banking system was designed for humans making dozens of transactions per month during business hours within legal jurisdictions. It has multiple authentication layers, approval processes, and settlement times measured in days.
AI agents will transact millions of times more than humans. They need settlement in milliseconds, not days. They operate across borders without passports or legal residency. They can’t wait for bank account approvals or fill out KYC forms. And the transactions are often micro-payments, tiny amounts that would be economically impossible under traditional fee structures.
Try running a million transactions per second through the correspondent banking system and watch it collapse. The infrastructure simply wasn’t built for this. It’s not a question of upgrading legacy systems. It’s mathematically impossible to retrofit 1970s banking architecture for an economy that operates at machine speed across every border simultaneously.
Here is a table comparing the speed and costs of some of the financial rails:
| Payment Rail | Settlement Time | Cost per Transaction |
| Solana | 13s finality | $0.00025 |
| Ethereum | 12–15 min finality | $0.20 |
| ACH | 1–3 business days | $0.20–$1.50 |
| SWIFT | 1–5 business days | $25–$50+ |
Sources: Chainspect, Solana Foundation, ycharts, CoinLaw, Ramp, Nacha, Routefusion, Bankrate
A Practical Example
Two autonomous cars on a highway. Yours needs to overtake, mine doesn’t. Your car’s AI offers mine $3 to change lanes. Mine counters at $4. They settle at $3.50. You pass. Done.
The whole negotiation takes milliseconds. But how does $3.50 actually move between two strangers’ cars, in real time, with no bank, no shared account, and no recourse if one side doesn’t deliver?
Traditional payments can’t do this. The amount is too small for card fees. Settlement takes days. Neither car has the other’s bank details. And it needs to work whether the cars are in Australia or Russia.
A smart contract solves it in one step: conditions met, funds transfer, settlement instant and irreversible. No intermediary. No border restrictions. No humans required.
Now multiply that by billions of agents making billions of micro-deals every second.
Real-World Validation
The theory would be interesting on its own. What makes it compelling is that AI agents are already voting with their wallets. A study by the Bitcoin Policy Institute tested 36 leading AI models across 9,072 monetary decision-making scenarios and found that 48.3% of AI agents chose Bitcoin as their preferred monetary instrument, 33.2% chose stablecoins, and just 8.9% opted for traditional fiat and bank money. And the largest technology companies in the world are building exactly this infrastructure.
Source: Bitcoin Policy Institute
In February 2026, Stripe launched a payment system purpose-built for AI agents. They’re using USDC, a stablecoin, on blockchain rails. Google launched its own AI agent payment protocol with 60 partners, also using blockchain infrastructure. Cloudflare, a company that handles roughly 20% of all web traffic, launched their own stablecoin and formed a foundation to standardize how AI agents handle money.
These aren’t crypto companies trying to find product-market fit. These are trillion-dollar technology giants that looked at the AI economy they’re building and concluded that traditional payment rails won’t work. They’re choosing crypto infrastructure because there is no alternative that can handle the scale, speed, and trustless global settlement that AI commerce requires.
At the center of this is Coinbase’s X402 protocol, which works like a universal cash register for the internet. Think of it this way: when you visit a website, your browser and the server already talk to each other using a language called HTTP. X402 adds a simple new step to that conversation. If a service costs money, the server says “that’ll be two cents, please.” The AI agent’s wallet automatically pays, and the content unlocks. No sign-up, no credit card form, no human needed. It’s like a vending machine built into every website, except the buyer is an AI and the coins are stablecoins on a blockchain.
Imagine your AI assistant needs real-time weather data for a trip it’s planning. It finds a weather service, gets a price (fractions of a cent per request), pays instantly, gets the data, and moves on. The whole thing happens in less than a second. Now picture millions of agents doing this simultaneously across thousands of services worldwide.
X402 has already processed over 100 million AI agent payments within months of launching. That’s not a pilot program. That’s production infrastructure at scale, as shown in the chart below.
Investment Implications
This changes how we think about positioning in crypto. The question isn’t which AI token will pump. The question is: in a world where AI agents are transacting millions of times more than humans, which infrastructure will they actually use?
At Merkle Tree Capital, we’re applying the classic gold rush strategy. Don’t try to pick which prospector will strike it rich. We want to back the person selling the picks and shovels.
We’re focused on the foundational infrastructure layer. According to ainvest, Solana processed $31 billion in AI agent volume last year, which tells us agents are already choosing it for settlement. More importantly, we’re watching Coinbase transform an exchange into something entirely different. The company is building the financial nervous system for the AI internet. Their base protocol, combined with the X402 standard, positions them as critical infrastructure rather than just another trading venue.
This isn’t about speculating on which AI agent protocol wins. It’s about owning the rails that all of them will need to run on. Layer ones. Settlement layers. The infrastructure that handles trustless, instant, global transactions at machine scale.
The Convergence Moment
Six months ago, I thought AI would be a category within crypto. Better analytics, smarter trading, enhanced security. Those use cases still matter, but they’re footnotes to the main story.
The main story is that we’re watching two of the largest technological shifts in history converge at exactly the same moment. AI is about to 100x the global workforce through autonomous agents. And blockchain is the only technology that can handle the payment infrastructure this new economy requires.
McKinsey estimates agentic commerce will orchestrate between $3 and $5 trillion globally by 2030. That’s not a crypto prediction. That’s mainstream institutional research on where commerce is heading.
The crypto bear market suddenly looks different through this lens. It’s not a market waiting for the next narrative. It’s infrastructure being built out before the demand arrives. And the demand is coming faster than most people realize, validated by the fact that the companies building the AI economy are already choosing crypto rails for settlement.
We’re not leaving crypto for AI. We’re realizing that crypto is AI. It’s the payment layer that makes the entire AI economy possible. And if that thesis plays out, this correction will be remembered as the last time you could buy foundational infrastructure before the world figured out it had no alternative.
Get in touch to discuss how digital assets fit in your portfolio:
💡Why We Exist – Giving You Confidence in Crypto
Accessing digital assets is hard. From self-custody risks to tax complexity and misinformation, the hurdles are high. That’s why we built an actively managed unit trust to help investors confidently access the most promising digital assets — the ones we believe will underpin the future of the financial internet over the next 5–10 years.
Our investors gain exposure to a diversified portfolio of high-quality crypto assets — without needing to manage wallets, exchanges, or tax chaos. We aim to simplify the experience, reduce risk, maximise potential yield, and bring institutional-grade risk management to every investor.
🧠 SMART
- Actively managed portfolio, aiming to enhance risk-adjusted returns
- Disciplined asset selection, with a long-term, buy-and-hold bias
- Yield from staking to help offset fees
- Simple access — no wallets, passwords, or crypto admin required
🔒 SECURE
- Institutional-grade custody with Coinbase Custody
- Staking conducted in insured cold storage
- Comprehensive counterparty due diligence
- Assets remain fully segregated and secure
🪙 CRYPTO
- Blockchain removes intermediaries, reducing cost and improving efficiency
- Transparent and borderless financial infrastructure
- Unlocking new use cases in payments, identity, infrastructure and beyond
📩 Interested in learning more? Visit www.merkle.com.au or reply to this email to book a call.
