Stripe launches Machine Payments Protocol to enable AI agents to transact autonomously

Why Stripe's Machine Payments Protocol Signals a Turning Point For Micropayments

For thirty years, micropayments failed for one reason: the payer was human. Humans hesitate. They abandon carts, resent friction, and perform mental accounting before every small transaction. The smartest engineers and best-funded companies in technology could not engineer their way around human psychology. So they stopped trying. On March 18, 2026, that constraint disappeared. Stripe and Tempo launched the Machine Payments Protocol -- and the entity doing the paying has no psychology to work around. The Micropayments Graveyard: Three Generations Of Failure To understand why MPP matters, start with why everything before it failed. Below is a condensed history of the most significant attempts -- organized by generation, each dying of a different cause. Figure 1. The 30-year history of micropayments. The pattern across all three generations is identical: every attempt assumed the paying entity was a person. MPP is the first protocol built for an entity that has no psychology to work around. Why MPP Is Structurally Different On March 18, 2026, Stripe and Tempo launched the Machine Payments Protocol alongside the Tempo mainnet -- a payments-focused blockchain co-developed with Paradigm. But one design principle separates it from everything above: MPP is built for agents to pay, not humans. Unlike every prior micropayment system, no human decision point exists between resource request and payment execution. Payment is a programmatic step, not a conscious decision. Figure 2. The MPP payment flow. This solves the thirty-year problem at its root. AI agents have no mental transaction cost. They do not hesitate before paying $0.001 for an API call. They do not abandon carts. They simply execute payment as a function call and continue with the task. Machine‑initiated payment models invert that assumption. They are designed for non‑human actors such as AI agents, automated workflows, and software services, that need to pay for resources such as data access, compute, APIs, or digital services in real time. From an architectural standpoint, protocols like MPP embed payment directly into the request‑response cycle between an agent and a service. Payment becomes a programmatic step, not a conscious decision. There is no "checkout moment," no cart abandonment risk, and no mental transaction cost. This is consistent with Forrester's broader research on agentic AI, where autonomous systems increasingly operate across procurement, finance operations, software development, and customer engagement. As agents take on more end‑to‑end responsibility, they require the ability to transact independently -- within predefined controls. Three structural shifts make this approach more viable than previous micropayment efforts: Unlike human consumers, agents cannot substitute free alternatives when a required service is unavailable. If a workflow depends on a specific dataset or API, payment is mandatory for task completion. This eliminates one of the biggest adoption barriers that undermined earlier models. Unlike past experiments that required new wallets, browser extensions, or proprietary networks, current approaches are designed to operate on top of existing card rails, bank transfers, and -- increasingly -- tokenized money. Initial demand is coming from enterprise and developer environments, not consumer content. This mirrors Forrester's predictions in B2B payments, where automation, reconciliation, and straight‑through processing drive willingness to pay -- even for small, frequent transactions. An Emerging Stack, Not a Single Winner One common misconception is that machine‑initiated payments will converge on a single dominant platform. Forrester's view is the opposite. The emerging ecosystem already spans multiple layers: * Protocol and standards, defining how agents request services and trigger payments. Besides Stripe's MPP, we also see Coinbase's x402 protocol, which Activates HTTP 402 for instant USDC payments directly over HTTP with Zero protocol fees, sub-2-second settlement and Crypto-native path. Two ex-Stripe engineers also founded Circuit & Chisel, which introduced Agent Transaction Protocol (ATXP). It helps AI agents navigate the web and execute payments as part of multi-step task completion. * Identity and trust frameworks, addressing the question of "who is this agent", a theme closely aligned with Forrester's research on AI agent identities management, zero trust, and fraud management for AI agents. Silicon Valley agentic payments startup Skyfire introduced Know Your Agent (KYA) protocol using JSON Web Tokens. Gives AI agents a verifiable identity so merchants can distinguish legitimate agents from malicious bots -- and accept payment from verified ones. KYAPay supports agent-to-agent, B2A, and A2B payments both on-chain and off. Skyfire recently built a strategic partnership with F5 to enable enterprises to securely allow verified AI agents while still blocking malicious actors. * Wallets and programmable controls, enabling spend limits, policy enforcement, and auditability. Blockchain wallet infrastructure provider Crossmint gives AI agents their own wallets, virtual Visa/Mastercard cards, and stablecoin accounts with programmable spending controls -- per-transaction limits, merchant whitelists, human approval thresholds. It is protocol-agnostic: supports both x402 and card rails. This mirrors the evolution of the internet itself: interoperable layers rather than a winner‑take‑all model. For banks and payment companies, this creates both opportunity and risk. Value will accrue unevenly across layers, and late entrants may find key control points already claimed. Implications For Banks And Payment Companies Financial institutions face a narrowing window. The institutions that wait for transaction volumes to materialize will find the identity, trust, and wallet layers already owned by fintechs, card networks, and crypto infrastructure providers. Three strategic options remain open -- for now: Conclusion Micropayments did not fail because the industry lacked imagination or engineering talent. They failed because the assumed payer -- a human -- was wrong. Machine-initiated payments remove that assumption permanently. The most important question for financial institutions is no longer whether this works. It is whether they move before the control points are claimed. That shift is already underway. Forrester clients can set up an inquiry or guidance session to discuss these topics with us. We will also publish new research around agentic payments soon. Stay tuned!

Stripe-backed crypto startup Tempo releases AI payments protocol, launches blockchain | Fortune

The fintech giant Stripe along with Tempo, a blockchain startup incubated by the payments company as well as the venture firm Paradigm, launched a new payments protocol on Wednesday to make it easier for AI actors to send and receive money. Dubbed the "Machine Payments Protocol," the open-source network supports payments in both fiat and cryptocurrency. The protocol is also compatible with Stripe's existing AI payments infrastructure. While the new AI payments network currently runs only on top of Tempo, which is also the name of the blockchain developed by the Stripe-backed startup, it's designed to operate across multiple blockchains and payments rails, not just one. On Wednesday, Tempo simultaneously announced that its blockchain went live after operating in a test phase over the past three-and-a-half months. "Agentic payments is very early, and we still are figuring out the best way to structure these," Matt Huang, cofounder of Tempo and managing partner at Paradigm, told Fortune. "So our team just came up with what we thought was the most elegant, minimal, efficient protocol that anyone can extend without our permission." Agentic payments has become the newest buzzphrase in fintech. The concept refers to when AI agents, or autonomous bots, send and receive money on behalf of their human users. The arena is nascent, but technologists envision a future where agents are crawling the web and paying each other to access news articles, buy products from Amazon, or download a dataset. Tempo, which raised $500 million at a $5 billion valuation in 2025 from Silicon Valley heavyweights like Joshua Kushner's Thrive Capital, has positioned itself to ride this new wave of commerce. The startup's blockchain is designed for high-speed payments and stablecoins, or cryptocurrencies pegged to real-world assets like the U.S. dollar. The protocol Tempo designed with Stripe isn't the only existing framework for agentic payments. Coinbase has also designed its own network, which it dubbed x402, a callback to a message encoded by early internet pioneers that returned a 402, or "Payment Required," error. And Google released a new payments scheme in September that includes support for credit cards as well as stablecoins. The payments giant Visa also contributed to Stripe and Tempo's "Machine Payments Protocol," developing the specifications for letting agents pay with credit or debit cards. "We look at MPP as another way that you can have a very clear, defined protocol around how an agent communicates with merchants," Cuy Sheffield, Visa's head of crypto, told Fortune.

Stripe-Backed Tempo Network Launches With Focus on AI Agent Payments - Decrypt

The standard, Machine Payments Protocol, has already been bolstered by Visa and others. Tempo, the payments-focused layer-1 blockchain from Stripe and Paradigm, launched its mainnet on Wednesday, providing key infrastructure for the agentic economy in the process. Alongside the network's launch, Tempo unveiled a new open standard for agentic payments -- the Machine Payments Protocol (MPP). The protocol, co-authored by Stripe and the Tempo team, circumvents the "limitations of existing payment rails," allowing agents to transact seamlessly. "MPP provides a standard way for agents and services to coordinate payments programmatically," the network wrote. "Instead of each service inventing its own billing flow, MPP defines a simple protocol for requesting, authorizing, and settling payments between machines." The open standard has already been enhanced by a pair of the network's early launch partners, including global payments giant Visa, which extended MPP to support card-based payments, and Lightspark, which did so for Bitcoin payments on the Lightning Network. Stripe too added on functionality, creating support for "cards, wallets, and other payment methods." "MPP lets an agent pay for services autonomously: An agent can request a resource from a service, and the service responds with a payment request," the announcement reads. "The agent then authorizes payment from its wallet, the transaction settles instantly, and the service delivers the requested resource to the agent." According to the network, the functionality is possible thanks to its introduction of "sessions," which allows for a stream of payments to be made programmatically based on existing, defined limits. The network has already released a directory of compatible services with which agents can interact with and pay for, including those from OpenAI, Anthropic, and Google. In addition to its agentic enhancements, Tempo says it "architected the infrastructure" to combat traditional, arcane solutions related to global payments, cross-border remittances, and more. The network's focus on the agentic economy amplifies a crescendoing trend among blockchain companies. In September, the Ethereum Foundation created its own artificial intelligence team, signaling the importance of the technology and its potential interplay with blockchain and cryptocurrencies. The foundation's initial focus was on ERC-8004, an Ethereum Improvement Proposal that similarly enables agents to transact seamlessly across the Ethereum blockchain. It also backed an open-source protocol from Google that tackles the same issue. Crypto exchange Coinbase has also been focused on enabling agentic payments, launching a wallet with built-in guardrails in February.

Stripe's crypto joint venture Tempo launches payments protocol for AI

Stripe describes Machine Payments Protocol as an 'internet-native way for agents to pay'. It has been decided by Big Tech bigwigs that consumers will be using AI to shop online, despite online shopper sentiment reflecting some level of apprehension. McKinsey has found that most European consumers already use AI to help shape their purchase decisions, but not at checkout, where money passes hands. Though it noted that that sentiment could change, and fast. McKinsey also found that by 2030, agentic commerce could orchestrate up to $5trn globally. But Morgan Stanley noted that that only 1pc of shoppers currently choose the agentic route, leaving much to speculation and a hope that consumers will let AI shop for them. Regardless, many, such as Revolut, Google and PayPal, are already trying to build protocols to support this new incoming age of AI-led shopping. Adding to that is Stripe's new joint venture with Paradigm, Tempo, which has launched the 'Machine Payments Protocol (MPP). Described as an "internet-native way for agents to pay", MPP is attempting to create an alternate financial system built specifically for AI agents to use. The protocol is offering a system where AI agents are not met with the challenges of navigating a financial system built for humans, where the bots might have to create shopping accounts, navigate pricing pages, choose between subscription tiers, enter payment details, and set up billing. "The tools of the current financial system were built for humans, so agents struggle to use them", noted Stripe's agentic commerce leads in a blog post. MPP lets agents and services coordinate payments programmatically, enabling micro-transactions and recurring payments. Once Stripe users set up MPP, businesses can accept payments directly from agents in both stablecoins and fiat, as well as use features such as buy now, pay later. Companies such as browser infrastructure provider Browserbase, and New York City-based Prospect Butcher already use MPP to allow agentic commerce. Tempo was reportedly valued at $5bn in October, after a $500m raise. The crypto venture launched from incubation a month prior. Last November, Swedish fintech giant Klarna, whose CEO was once a crypto sceptic, became the first bank to launch a stablecoin on Tempo called 'KlarnaUSD'. The coin, set to launch this year, is expected to help cut down on transaction fees. At the time, sources told the Financial Times that the stablecoin would also help Klarna move large amounts of money globally by cutting out parties such as the Society for Worldwide Interbank Financial Telecommunications - or SWIFT - network. Meanwhile, Stripe acquired US billing and invoice software provider Metronome in December. Metronome enables organisations to create and manage user-based pricing models, which Stripe CEO Patrick Collison said is "the native business model for the AI era". Bloomberg news reported last month that Stripe was considered acquiring some of all parts of its long-standing fintech rival PayPal. Founded in the late 1990s, PayPal has struggled in recent years to modernise against its emerging rivals. The 2010-founded Stripe, meanwhile, was recently valued at $159bn after an employee tender offer. Don't miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic's digest of need-to-know sci-tech news.

Stripe Protocol Could Revive Micropayments With AI Agents: Forrester

Forrester says Stripe's Machine Payments Protocol reflects a shift toward automated transactions, as AI agents remove behavioral barriers that hindered micropayments. Stripe's newly launched Machine Payments Protocol (MPP) could mark a turning point for micropayments -- a long-promised but underutilized use case in crypto and beyond -- as AI agents reshape how transactions are made. That's the key takeaway from newly published analysis by Forrester senior analyst Meng Liu, who argues that MPP may succeed where decades of earlier efforts failed. Introduced earlier this month, MPP enables AI agents to execute transactions automatically, removing the need for human approval at each step. It is described as an open protocol for coordinating payments between AI agents and services. Liu frames this as a structural shift from human-initiated payments to machine-to-machine transactions. Micropayments, which are typically small transactions worth a few cents or dollars, have long been seen as a way to monetize digital content, services and data, but have struggled to gain traction at scale. A major barrier to adoption has been human behavior, including cumbersome digital checkout processes and reluctance to approve small charges, Liu said. By contrast, AI agents executing payments as part of task completion, such as paying to access data or use online services, eliminate those constraints. "Payment becomes a programmatic step, not a discrete decision," Liu wrote. "There's no checkout moment, no cart abandonment risk, and no mental transaction cost." Importantly, MPP is not a new settlement network. Instead, it acts as a coordination layer for automated payments, designed to work across existing infrastructure, including traditional rails, digital wallets and, where supported, crypto rails. Related: AI agent payment volumes lower than reported, but adoption is growing: a16z Stripe is a payments company that has expanded into digital assets, including support for stablecoins, crypto on-ramps and blockchain-based payment tools. While MPP itself is not inherently blockchain-based, other companies are also developing infrastructure for AI-driven payments, particularly in areas such as micropayments and autonomous transactions. One recent example is MoonPay, which released an open-source wallet standard designed for AI agents. The framework allows agents to hold, send and receive digital assets, enabling them to transact independently without human intervention. Meanwhile, analysts at Bernstein believe AI agents could boost demand for stablecoins, as they are well-suited for handling frequent, low-value payments. Like Forrester's Liu, Bernstein also pointed to Coinbase's x402 protocol, which enables automatic internet payments between machines.

AI Agents Get New Tools From Visa and Stripe's Tempo

AI agent users received new tools from Visa and the Stripe-backed Tempo to provide a new way for agentic payments to take place online. Visa's crypto division has launched a tool to allow artificial intelligence agents to make payments, the same day the Stripe-backed blockchain Tempo launched alongside a protocol for AI agents. "Excited to share Visa CLI, the first experimental product from Visa Crypto Labs," Cuy Sheffield, the head of Visa Crypto Labs, posted to X on Wednesday. A website for Visa CLI, meaning a command line interface where users type what action a program must take, says the tool will give an AI agent "the ability to securely pay for what you need as you code." The tool also said it allows for "programmatic card payments without the pain of API keys." API keys can include sensitive information that AI agents can leak, causing security risks. It's the latest standard seeking to allow AI agents to make payments online as hype around AI and stablecoins grows. Coinbase launched its x402 standard to facilitate agentic stablecoin payments in May, which was most recently integrated by Sam Altman's World in a developer toolkit for AI agents released on Tuesday. Meanwhile, the Tempo blockchain, backed by payments company Stripe, launched on mainnet on Wednesday, releasing a payments protocol for AI agents. Tempo posted to X that its blockchain was "purpose-built for payments" and focused on servicing high-throughput stablecoin transactions, currently one of the most popular ways AI agents are used. "Agents can already write code, coordinate services, retrieve data, and execute complex workflows across the internet. But as these systems become more capable, they increasingly need to transact," Tempo said. The project also launched the Machine Payments Protocol, an open standard that it developed with Stripe, which it described as giving "a standard way for agents and services to coordinate payments programmatically." Related: SlowMist introduces Web3 security stack for autonomous AI agents Tempo said the protocol "is designed to be rail-agnostic and extensible," noting that Visa had extended support for the protocol on its card payments network while Stripe is supporting "cards, wallets, and other payment methods." The crypto fintech Lightspark had also extended support for the protocol over the Lightning Network for Bitcoin (BTC) payments.

Stripe-Backed Protocol Lets AI Agents Transact Autonomously | PYMNTS.com

By completing this form, you agree to receive marketing communications from PYMNTS and to the sharing of your information with our sponsor, if applicable, in accordance with our Privacy Policy and Terms and Conditions. With the mainnet now live, developers can now build on Tempo through its public remote procedure call (RPC) endpoints, Tempo said in a Wednesday blog post. "Tempo is infrastructure for real-world payments at internet scale," the company said in the post. "It's designed for instant settlement, predictable low fees, high throughput and global availability." In the time since Tempo was announced in September 2025, the rise of agentic AI has made clear the need for agentic payments for things like access to a dataset, compute or testing infrastructure, and services needed to complete tasks, according to the post. "Tempo provides the settlement infrastructure for this scale of interactions, allowing agents to transact programmatically," Tempo said in the post. The infrastructure also supports established payment flows such as global payouts, cross-border remittances, embedded finance and tokenized deposits, per the post. The open standard for machine payments announced Wednesday by Tempo is called the Machine Payments Protocol (MPP) and is co-authored by Stripe and Tempo, according to the post. MPP provides a standard way for agents and services to coordinate payments programmatically and enables machine payments to work across services and payment rails, per the post. It currently works with stablecoins, cards and other supported payment methods. "MPP runs on Tempo today, but the protocol itself is designed to be rail-agnostic and extensible," Tempo said in the post. "For example, our design partner Visa has already extended MPP to support card-based payments on their network." When Tempo was introduced in September 2025, the company said its payments-first blockchain was purpose-built for stablecoins and real-world payments. The blockchain's priorities include predictable low fees, opt-in privacy, payments/gas in any stablecoin, a payments-first user experience, and scale, the company said at the time. The company launched its public testnet in December 2025, saying that anyone could start building on Tempo and that the network was being tested by many of the world's leading companies. Matt Huang, co-founder and managing partner at Paradigm and project lead for Tempo, told Bloomberg in December: "The crypto ecosystem can be quite intimidating. We want to close that developer experience gap for people thinking about real-world use cases for stablecoins."