For years, digital payments have been designed around human behaviour: create an account, enter card details, approve the purchase, and come back when you need the service again. That model works reasonably well for people, but it creates friction for AI agents. If software is going to plan tasks, call tools, buy data, and complete workflows autonomously, it also needs a simpler way to pay. Stripe’s Machine Payments Protocol (MPP) is one of the clearest signs yet that payment infrastructure is starting to evolve for that reality.
What Stripe MPP actually is
MPP is an open, internet-native protocol co-authored by Stripe and Tempo for machine-to-machine payments. At a high level, it allows an agent to request a paid resource from an API, service, MCP endpoint, or other HTTP-accessible system, receive a payment challenge, authorise the payment programmatically, and then retry the request with proof of payment. Instead of forcing agents through a human checkout flow, the payment becomes part of the request-response cycle itself.
Why this matters now
The rise of agentic systems is changing the economics of software. A human buyer might purchase a monthly subscription and use a tool a few times a week. An agent, by contrast, might need to make thousands of tiny purchasing decisions every day: paying for browser sessions, premium API calls, data lookups, document processing, specialist tools, or one-off task execution. That makes per-request commerce far more important than traditional seat-based pricing. MPP is designed for exactly that environment, where software becomes both the user and the buyer.
How the protocol works in practice
The flow is intentionally simple. An agent asks for a resource. If the resource requires payment, the server responds with an HTTP 402 payment-required message containing the payment details. The agent then authorises the payment and retries the request. Once payment is confirmed, the server returns the resource together with a receipt. From a product strategy perspective, that is important because it removes the need to bolt a separate commercial process onto every automated interaction. Payment becomes a native capability of the endpoint itself.
The commercial opportunity for businesses
For businesses selling digital capabilities, MPP opens up new pricing models that were previously awkward or uneconomical to implement. Rather than forcing every prospect through account creation and subscription selection, companies can let agents pay per invocation, per session, or per outcome. That is especially attractive for APIs, premium data products, document workflows, infrastructure services, and specialist tools where usage is episodic or highly variable. Stripe has already highlighted examples such as Browserbase charging for browser sessions and PostalForm enabling agents to pay to print and send physical mail.
Why Stripe’s implementation is strategically interesting
What makes Stripe’s move notable is not just the protocol itself, but the surrounding infrastructure. Stripe positions machine payments so they can flow into the same operational systems businesses already use for conventional payments: balances, reporting, refunds, payout schedules, and broader financial operations. MPP also supports two payment paths. For lower-value machine transactions, businesses can accept stablecoin-based payments. For more traditional payment methods, Stripe ties MPP into Shared Payment Tokens (SPTs), which allow an agent to pass limited-use payment credentials with amount and expiry controls. That matters because it bridges autonomous commerce with familiar payment rails rather than treating machine payments as a completely separate stack.
What business leaders should watch carefully
MPP is promising, but it should be viewed as an emerging infrastructure layer rather than a finished mass-market standard. Today, Stripe documents machine payments as a private preview, and there are still regional and operational considerations around stablecoin acceptance and rollout. For most organisations, the immediate question is not whether every product should suddenly support MPP. The better question is whether there are high-value, machine-consumable interactions in your business that would benefit from pay-per-use access. If the answer is yes, then MPP is worth monitoring closely and potentially piloting.
A practical lens for evaluation
Businesses are most likely to benefit when three conditions are true. First, the product can be consumed programmatically through an API, tool, or service endpoint. Second, usage is variable enough that rigid subscriptions create friction. Third, the buyer increasingly looks like an agent rather than a person. When those factors align, machine-native pricing can become a growth lever rather than just a technical novelty.
Our view
Stripe’s Machine Payments Protocol is important because it treats payments as infrastructure for autonomous systems, not just checkout for humans. Whether MPP becomes the dominant standard or one of several competing approaches, the direction of travel is clear: agents are becoming economic actors. The organisations that adapt early will be the ones that make their services easier for software to discover, buy, and use. Over the next 12 to 24 months, that may become a meaningful competitive advantage for digital businesses building in the AI economy.
What is Stripe's Machine Payments Protocol (MPP)?
Stripe MPP is a protocol that enables AI agents to autonomously initiate and authorise payments for services and APIs on behalf of users or businesses. It extends Stripe's payment infrastructure to support agentic commerce — transactions initiated by software rather than humans.
How does agentic commerce differ from traditional e-commerce?
In traditional e-commerce, a human reviews and approves each transaction. In agentic commerce, an AI agent evaluates options, selects a service, and completes payment autonomously within pre-defined parameters — enabling real-time, programmatic procurement at machine speed.
Is Stripe MPP available to UK businesses now?
Stripe MPP was announced in 2025 and is rolling out progressively. UK businesses with existing Stripe integrations are well positioned to adopt MPP as it becomes generally available, given Stripe's strong presence in the UK market.
What safeguards prevent AI agents from making unauthorised payments?
MPP is designed with spend limits, merchant category restrictions, and approval workflows that businesses configure in advance. Agents operate within explicitly granted permissions — they cannot exceed defined budgets or transact with unauthorised vendors.
What types of AI agent use cases does Stripe MPP enable?
MPP unlocks use cases such as agents automatically purchasing API credits when a quota is reached, procuring research data on demand, booking services as part of a workflow, or settling inter-agent transactions in multi-agent systems.
How does Stripe MPP handle fraud and dispute resolution for agent-initiated payments?
Stripe's existing fraud detection infrastructure applies to MPP transactions, supplemented by agent-specific signals such as transaction velocity and behavioural anomalies. Dispute resolution follows standard Stripe processes, with transaction logs providing detailed audit trails.
What are the accounting and compliance implications of AI-initiated payments?
Businesses must ensure agentic payments are captured in their accounting systems with appropriate categorisation and approval records. Finance teams should define MPP payment policies that align with existing procurement controls and VAT treatment requirements.
How should businesses prepare their systems for agentic payment integration?
Start by identifying which agent workflows currently require human payment approval, then assess whether the value and risk profile suits automation. Implement spend controls, logging, and alerting before enabling autonomous payment for any production workflow.
Will Stripe MPP work with AI platforms like OpenAI, Anthropic, and others?
Stripe is building MPP as an open protocol designed for interoperability across AI platforms. Early integrations focus on major model providers, with broader ecosystem support expected as the protocol matures.
What does the rise of machine payments mean for the future of B2B commerce?
Machine payments will compress procurement cycles from days to seconds for routine purchases, shift pricing models towards usage-based and API-native structures, and require finance teams to govern a new category of autonomous expenditure. Businesses that design for this shift early will gain significant operational advantages.
Get in touch today
Book a call at a time to suit you, or fill out our enquiry form or get in touch using the contact details below