What Is the Machine Payments Protocol — and How Will Agents Pay for Services?
# What Is the Machine Payments Protocol — and How Will Agents Pay for Services?
The Machine Payments Protocol (MPP) is an open standard—a shared “billing grammar”—that lets AI agents and other automated software request, authorize, and settle payments programmatically, without a human stepping through checkout each time. Co-authored by Stripe and Tempo, MPP launched on March 18, 2026 alongside the Tempo mainnet, and it’s designed to be payment-rail agnostic, so machine-to-machine commerce can work across everything from stablecoins to card networks.
The problem MPP is trying to solve
As AI agents move from demos to doing real work, they immediately hit a practical wall: the internet’s paid services—APIs, datasets, compute, and model inference—typically assume a human is present to sign up, confirm a purchase, and manage billing. In other words, an agent can “decide” to use a tool, but it can’t reliably pay for it in a standardized way.
MPP’s core goal is to replace today’s patchwork of bespoke billing integrations with a common protocol layer that services and agents can both implement. If a service speaks MPP, an agent can discover the terms, authorize payment, and receive confirmation in a consistent format—rather than every provider inventing a custom paywall.
How MPP works: a typical transaction flow
At a high level, MPP sits between an agent and a service to standardize the back-and-forth required to pay for and unlock a digital resource.
A typical MPP flow looks like this:
- The service advertises price and payment terms.
Think of this as the service declaring, in a machine-readable way, what it costs to access a resource (an API call, a dataset query, browsing time, etc.) and under what conditions.
- The agent requests the resource.
The agent attempts to use the service, and the service responds with the information needed to pay.
- The service returns a payment request.
Instead of redirecting to a human checkout, the service produces a structured request the agent can process.
- The agent authorizes payment from a wallet or session balance.
Authorization is the key: the agent needs explicit permission and a mechanism to spend. MPP supports a session mechanism that’s been compared to “OAuth in payments”—meaning the agent can get a one-time authorization and then transact repeatedly under defined constraints.
- Settlement completes (ideally instantly), and the service delivers the resource.
Once the payment is confirmed, access is granted and the agent can continue its workflow.
The session concept matters because high-frequency agentic workflows can’t stop for manual approval every time—and they also can’t necessarily afford the overhead of a separate, heavyweight payment step for every single call. MPP’s sessions are intended to enable ongoing real-time settlement once an agent has been authorized (for example through pre-funding or a credit-backed session), rather than treating each interaction as a fresh checkout.
Payment modes and infrastructure MPP supports
MPP is explicitly designed to be rail-agnostic and extensible, but it emphasizes a few key modes:
- Stablecoins as a foundational rail for instant, cross-border, 24/7 settlement. The premise is that agents operating globally (and continuously) benefit from payment instruments that can settle at internet speed, without banking-hour constraints.
- Shared Payment Tokens (SPTs) to bridge into fiat-based systems such as credit/debit cards and buy-now-pay-later (BNPL). In the ecosystem described around MPP, Visa contributed specifications for card-based agent payments—an indicator that traditional rails are being considered alongside crypto-native ones.
- Tempo mainnet as an early deployment rail, positioned around high throughput, low fees, and predictable costs—attributes that matter if agents start making many small payments (or very frequent ones). Importantly, MPP “currently runs on Tempo” in early deployments, while still aiming to be a standard that other rails can implement.
On the merchant and platform side, Stripe integrated MPP into its PaymentIntents API, aligning MPP flows with familiar merchant tooling and operational needs such as taxes, reporting, and fraud protection. In practice, that signals a strategy: make agent payments look less like an exotic new payment type, and more like something businesses can adopt without rebuilding their entire billing stack.
(If you’re tracking the business angle: TechScan has also covered the broader push toward standardizing agent micropayments in Stripe Pushes Standard for AI Agent Micropayments.)
Why MPP could change agentic commerce
MPP’s benefits are straightforward—but compounding:
- Autonomous consumption at scale. Agents can discover and pay for services as part of an end-to-end task, instead of failing at the “and then pay” step.
- Less fragmentation across paid APIs and tools. A standardized protocol means providers don’t each build a one-off “agent billing” system, and agents don’t need a custom integration per vendor.
- Micropayments become more realistic. MPP is designed around instant settlement and predictable low fees, which is critical if the unit of value becomes “per call,” “per page,” or “per query”—not monthly subscriptions.
Those traits unlock business models that map naturally to agent behavior: pay-per-inference, pay-as-you-go data access, and automated procurement across many services—without stitching together dozens of incompatible billing flows.
Why It Matters Now
MPP’s timing is tied to a visible shift: agents are moving from novelty to utility, and utility runs into paid services. The March 18, 2026 launch—with both Stripe and Tempo attached—puts a concrete, open standard into the ecosystem right as more agent workflows are expected to touch real commercial systems.
It also matters because participation isn’t limited to a single vendor stack. The brief reports contributions or integrations involving major players including Visa, Mastercard, OpenAI, Anthropic, and Shopify, alongside Stripe and Tempo. Even if adoption is gradual, that breadth increases the odds that MPP becomes an interoperability layer rather than a niche protocol.
If the last era of developer platforms was about making APIs easy to call, this moment is about making APIs easy to pay for automatically—and doing it in a way that works across rails.
Ecosystem and early use cases
MPP isn’t just theory: the ecosystem examples highlight agent-driven workflows already using it to pay for services such as:
- Browserbase (agent-driven browsing),
- PostalForm (form submissions),
- Prospect Butcher (lead enrichment).
On the acceptance side, Stripe’s MPP support inside PaymentIntents suggests a path for existing merchants: accept agent payments while keeping familiar controls and operational tooling.
Tempo, meanwhile, provides the early blockchain rail optimized for the high-frequency, low-fee settlement profile these interactions may demand—while the standard remains intended to be extensible so other rails can “plug in.”
Open questions and regulatory considerations
Even with a clean protocol, machine payments raise hard real-world issues:
- Stablecoin dependence and compliance complexity. Cross-border, always-on settlement is attractive, but custody, operational reliability, and jurisdiction-specific compliance can complicate deployments.
- Standardization requires coordination. Many providers must implement MPP and align on details like session semantics, how SPTs map to fiat rails, and what “confirmation” means across different settlement layers.
- Risk controls and auditability. As agents gain autonomous spending ability, systems need strong fraud mitigation, clear authorization boundaries, and robust audit trails—especially when payments happen at high frequency.
What to Watch
- Adoption signals: more major platform integrations (for example across model providers, commerce platforms, or developer toolchains).
- Cross-rail maturity: real end-to-end flows using SPTs with card rails or BNPL, not just crypto-native settlement.
- Operational + regulatory shifts: changes affecting stablecoins and automated payments that influence where and how MPP-backed settlement can run.
- New billing patterns: experiments that move beyond subscriptions into micropayments per inference, per-page browsing fees, or hybrid subscription+usage structures.
Sources:
https://www.techflowpost.com/en-US/article/30754
https://coincentral.com/stripe-and-tempo-build-high-speed-blockchain-for-ai-payment/
https://fortune.com/2026/03/18/stripe-tempo-paradigm-mpp-ai-payments-protocol/
About the Author
yrzhe
AI Product Thinker & Builder. Curating and analyzing tech news at TechScan AI. Follow @yrzhe_top on X for daily tech insights and commentary.