M2M Economics: AI Agents and the Future of Machine Payments

Artificial Intelligence (AI) has pervaded more and more businesses across different sectors. Data shows 88% of organizations are using AI, up from 50% in 2022, with the technology projected to contribute $15.7 trillion to the global economy by 2030.

As adoption accelerates, the role of systems powered by advanced machine learning has shifted from narrow, assistive roles to deep embedding in commerce, finance, logistics, and other industries.

These AI systems help improve efficiency, reduce costs, and enable automation by making informed decisions and optimizing workflows in real-time. They, however, depend on humans. This dependence is now evolving to greater autonomy.

As both the technology and its adoption continue to mature, a profound shift is emerging. AI isn’t just a tool anymore; it is becoming an economic actor, giving rise to Machine-to-Machine (M2M) Economics. At the center of this shift are autonomous software entities called AI “agents,” which operate within an underlying architecture comprising a few key elements.

At its core is a model, which helps the agent understand the intent and then provides the sequence of steps it can take across apps and systems to achieve the desired outcome.

Then, carrying out the steps requires secure integrations with inventory tools, logistics platforms, and payment gateways so the agent can perform actions such as placing an order, scheduling deliveries, or issuing refunds. Supporting this is a customer data platform (CDP), which stores user preferences, service history, and consent, along with knowledge graphs that map relationships between products and policies, ensuring agents operate with context rather than guesswork.

Over time, reinforcement learning from interactions such as clicks, conversions, and complaints helps agents learn which actions lead to better outcomes, making the experience more dependable.

Together, these layers form a coordinated system that enables successful agentic commerce.

AI agents don’t just assist, though; they also make decisions, execute tasks, and interact with other systems, with minimal to absolutely no human supervision. More striking, these proactive, outcome-driven agents can now buy and sell products and services from one another. An AI model automatically pays a cloud provider for compute resources. Another negotiates API access. They allocate resources and optimize costs autonomously.

According to Nvidia CEO Jensen Huang, these AI agents could create a “multi-trillion-dollar opportunity” for many industries, calling them “the new digital workforce.”

“The age of AI Agentics is here,” announced Huang at the 2025 CES.

This can be seen with retail giants like Walmart using AI agents to automate personal shopping experiences and facilitate merchandise planning. Even banking giant JPMorgan Chase is exploring them to detect fraud, provide financial advice, and automate legal and compliance processes.

Meanwhile, Google, Microsoft, IBM, and Salesforce have embedded agentic AI capabilities directly in their software platforms.

What we are currently witnessing is the early formation of an economy where machines aren’t just participants but autonomous buyers and sellers, working on behalf of a user or a system.

AI as Economic Actors: Buying and Selling with Digital Money

As AI agents begin to transact, the question is, how are they paying? Traditional financial systems do not work for these machines as they are built around humans. These systems require manual approvals, physical presence, human identity verification standards, and set banking hours that simply don’t work for autonomous, real-time machine interactions.

What AI agents need is a programmable, globally interoperable, and always-available payment system. This is where digital currencies come into the picture.

Stablecoins, a type of cryptocurrency designed to maintain a steady value by pegging their price to an asset like the US dollar, are uniquely suited for this machine-to-machine commerce.

For starters, they are programmable, which introduces intelligence into transactions and exactly what AI agents need. What this means is that logic, rules, and conditions can be put directly into the movement of funds.

By allowing payments to be executed automatically through smart contracts, stablecoins enable AI agents to transact conditionally, for instance, releasing funds automatically once a service is delivered or adjusting spending based on real-time conditions, something traditional banking systems struggle to support.

Stablecoins like USDT and USDC also provide instant settlement. Unlike legacy systems that may take hours or days, blockchain-based transactions settle in seconds, if not minutes. For AI agents operating in real time, this speed is essential.

More importantly, they are borderless, and with AI agents operating without national boundaries, stablecoins enable smooth, seamless cross-border fund movement by eliminating currency conversion and regulatory delays.

Then there’s the fact that the permissionless blockchain networks are always on. There are no weekends or public holidays that restrict the flow of funds in the crypto realm. By operating 24/7, stablecoins align perfectly with autonomous systems that do not follow business hours either.

These benefits enable microtransactions, allowing AI agents to pay as little as fractions of a cent, thus unlocking new business models that were simply not possible due to payment friction. When dealing at the sub-dollar level, traditional payment economics don’t make much sense, as banks and payment processors like Visa and Mastercard charge a fixed fee per transaction, making the transfer of small amounts economically unviable.

Hence, Circle, the issuer of the second-largest stablecoin USDC, is currently building its own payment systems designed specifically for machine commerce. Just last month, the company launched nanopayments, allowing agents to send as little as $0.000001 in USDC with no fee on its new Arc blockchain.

Stablecoins are being widely adopted. In 2025, total stablecoin transaction volumes surged to $33 trillion, and adoption extended beyond crypto traders. A report found that 33% of global finance leaders already use stablecoins in their business operations, with a whopping 86% interested in doing so in the next couple of years.

So, using “digital dollars”, AI agents can transact near-instantly, round the clock, globally.

“Very soon, there are going to be more AI agents than humans making transactions,” announced Coinbase CEO Brian Armstrong in a X post, citing their inability to open a bank account. “But they can own a crypto wallet,” he added.

And that’s why the crypto exchange has been moving “to an AI-first mentality throughout the company,” as shared by Armstrong recently. This includes Coinbase-incubated Base, where agents operate on-chain like mini-businesses, working to make the L2 agent-native.

Then there’s crypto venture capital firm Paradigm, which partnered with mainstream financial company Stripe to launch a payments-focused blockchain called Tempo, which will handle global payouts, microtransactions, and AI-agentic payments with its stablecoin-first design.

All these developments are forming a significant new economic layer, one that McKinsey projects could mediate $3-$5 trillion in consumer commerce by 2030.

AI Agents as a New Class of Customer

The advancement of AI technology and the expansion of M2M economics are leading to the emergence of AI agents as a new type of customer.

Unlike human consumers, these agents operate continuously, without tiring or getting bored. Also, they can go on forever, that too without the influence of sentiments and feelings. Emotions are not the basis of decision-making here; rather, it is based on data and optimization.

Also, this new class of customer prioritizes cost and efficiency above all else and can scale instantly. AI agents can go from handling a few tasks to massive volumes very quickly and with little extra effort. And all these factors will fundamentally change demand patterns in the economy.

Up until now, humans have been the main focus of businesses, but not anymore. They are now beginning to design products and services specifically for the consumption of machines i.e., APIs, data feeds, and compute services.

And that means, instead of marketing to humans, companies will optimize their offerings for algorithmic selection, ensuring that their services are the fastest, cheapest, or most reliable option for the AI agents making the decisions.

“This changes a lot about how we think about the investment landscape and about building products. You really have to think agent-first now and assume that most of your customers are going to be agents rather than people.”

– Matt Huang, managing partner at Paradigm

In this context, the role of programmable digital payments becomes even more central. And with AI agents needing fast, cheap, programmable money to transact, the stablecoin industry has found its use case.

According to Dan Morehead, the founder of Pantera Capital, which manages $5 billion in digital assets, AI has no choice but to use crypto for financial transactions because AI agents cannot handle physical cash. And no matter which company ultimately wins the AI race, they will all need blockchain infrastructure to move money, he believes.

Though stablecoins show promise in enabling agentic commerce, they still face challenges.

“Most people don’t have a stablecoin wallet,” Rubail Birwadker, Visa’s global head of growth, told Bloomberg. This gap between narrative momentum and infrastructure reality matters deeply for agents operating at scale. “It’s incredibly important that we separate the very well-deserved attention that stablecoins get in the agentic world, and the reality,” Birwadker added.

Infrastructure gaps are only part of the challenge. Regulatory uncertainty looms equally large. The legal status of an AI agent remains undefined, and questions about liability, such as who’s responsible when an agent breaks the law or makes an error, remain unanswered.

Also, when the entity initiating and handling the transaction is a program rather than a human, compliance with Anti-Money Laundering (AML) regulations becomes difficult.

But when it comes to stablecoins, they have begun to gain clarity, with the Markets in Crypto-Assets (MiCA) framework in the EU demanding strict reserve requirements for stablecoin issuers. In the US, the GENIUS Act has established a federal regulatory framework for U.S. dollar-pegged stablecoins.

These developments are enabling regulatory compliance for crypto firms and massive adoption in payments, helping stablecoins shift from a niche asset to mainstream financial infrastructure.

The Technology Enabling Autonomous Contracts and Payments

As AI agents become the new big class of customers, the question isn’t just about how they pay but also about infrastructure. For AI agents to fully participate in the economy, they must be able to transact legally and securely, which requires new technical and legal standards.

One of the key innovations enabling this is a set of protocols that allow AI agents to discover services, negotiate terms, and execute payments.

The infrastructure layer is currently formed by a few key emerging protocols, including Anthropic’s Model Context Protocol (MCP) and OpenAI’s Agentic Commerce Protocol (ACP).

In April last year, Google also introduced the Agent-to-Agent (A2A) protocol, a communication protocol for AI agents. Then, earlier this year, the tech giant announced the launch of the Universal Commerce Protocol (UCP) to create a unified system across the shopping experience, so retailers don’t have to build their own tools.

Card networks are also busy positioning themselves to own this agentic shift through Visa Intelligent Commerce and Mastercard’s Agent Pay. Meanwhile, crypto payments made by AI agents are flowing through x402, an open standard developed by Coinbase to provide online service providers a way to charge them directly.

The thing is, today, even simple tasks like renting computing power require signing up for services, providing your card details, and generating an API key so that the software can access another service. This whole setup becomes even messier the more sophisticated and ambitious a project.

But with x402, Erik Reppel, creator of the standard and head of engineering at Coinbase Developer Platform, points out that, “your wallet becomes the universal API key that lets you access any x402-enabled service.”

X402 offers a pay-per-use model, where when an agent requests a service, the server responds with a price, and the agent pays it automatically in crypto from a wallet assigned by its developer.

“Tools like x402 enable agents to pay for data, APIs, and services… It’s still early, but the momentum is real—x402 is unlocking a whole new class of apps and agents that can pay as they go,” said Reppel.

Launched about a year ago, AI assistants have since made more than 100 million transactions through the standard. Over the past 30 days, nearly 1 million transactions have taken place on X402, generating $2.5 million in volume, according to data provider Artemis.

About 4,000 merchants, including Amazon Web Services (AWS), Messari, Pinata, Heurist, and Alchemy, are willing to sell through the standard. But payments alone aren’t enough. For AI agents to operate as fully autonomous economic actors, they also need to sign contracts and establish legal identity. For an AI to sign a contract, it must have a unique, tamper-proof identity to establish provenance as to who signed it. This is where Decentralized Identifiers (DIDs) and Verifiable Credentials (VCs) come in.

In DIDs, AI agents are assigned a unique identity number that’s anchored to a blockchain, allowing them to sign transactions, authenticate themselves to services, and establish trust without human intervention by using a private key they control. Meanwhile, VCs are digital attestations issued to the AI agent, allowing them to prove their authorized capabilities without revealing private data.

Then there are smart contracts, self-executing code that automatically executes agreements when predetermined terms and conditions are met. They allow AI to enter into contracts where payment is released only upon delivery of a service.

This stack of identity, payments, and execution layers is enabling AI agents to operate end-to-end without human oversight.

The Stock to Watch

In the realm of AI agents, Teradyne (TER ) stands out for building the critical physical layer while much of the world focuses on software and digital payments.

The company is building robots that AI agents can control in real-world environments. For an AI agent to move a box in a warehouse, it needs a robot arm, and Teradyne is the world leader in these collaborative robot platforms through Universal Robots and Mobile Industrial Robots (MiR)

Teradyne’s robots can be controlled programmatically, making them ideal endpoints for AI agents operating in the physical world. More importantly, Cobots are easier to deploy than traditional industrial robots, enabling faster adoption across industries.

Its strong AI capabilities have sent its shares surging to new highs. As of writing, Teradyne shares are trading at $369 that puts its market cap at $57.6 billion. TER is up over 90% YTD and 399.65% over the past year.

The company has an EPS (TTM) of 3.48 and a P/E (TTM) of 105.65. It pays a dividend yield of 0.14%.

For the fourth quarter, Teradyne reported a 44% increase in revenue from a year ago to $1.08 billion. AI drove more than 60% of that revenue, with the company expecting this share to rise above 70% in the current quarter. CEO Greg Smith attributed the momentum to data centers being “the prime mover of the market,” with the company guiding for year-over-year growth in 2026 driven by AI demand in compute.

Segment-wise, Semiconductor Test, which includes operations related to the design and manufacturing of semiconductor test products and services, accounts for the most revenue at $883 million. The Product Test segment, which covers products and services for defense and aerospace testing, silicon photonics testing, wireless test systems, and circuit-board testing, recorded $110 million in revenue, while the Robotics segment, which includes autonomous mobile robots and collaborative robotic arms, reported $89 million.

“Our Q4 results were above the high end of our guidance range, fueled by AI-related demand in compute, networking, and memory within our Semi Test business,” said Smith, noting “sequential growth” across all of the segments and achieving a 13% growth at the company level in 2025.

For this period, Teradyne also reported adjusted earnings of $1.80 per share. Net income for Q4 was $257.2 million, or $1.63 per diluted share, up from $146.3 million, or 90 cents per share a year ago.

The robotics company also issued strong guidance, expecting “year-over-year growth across all of our businesses, with strong momentum in compute driven by AI.” It is forecasting first-quarter revenue between $1.15 billion and $1.25 billion and adjusted earnings per share in the $1.89 to $2.25 range.

Latest Teradyne, Inc. (TER) Stock News and Developments

Conclusion

As AI adoption across industries surges, it is redefining the way work is done by enhancing efficiency and reducing costs. More profoundly, intelligent systems are evolving from tools into autonomous agents, now reshaping who participates in the economy as well.

The ability of these AI agents to transact is driving the emergence of machine-to-machine payments, powered by digital currencies such as stablecoin. This is resulting in a shift, where payments become instantaneous, programmable, and embedded directly into software logic, enabling a new class of economic activity that operates continuously and globally.

Looking ahead, the convergence of AI, digital payments, and robotics will define the next phase of this transformation. AI agents will not only make decisions and move money but also increasingly control physical systems, execute tasks in the real world, and reshape entire industries.