Paradigm and Stripe launch Tempo mainnet: a payments-first blockchain for the agentic web
Mar 18, 2026 with Georgios Konstantopoulos
Key Points
- Paradigm and Stripe launch Tempo mainnet with a Machine Payments Protocol that lets AI agents autonomously pay for API calls, already available as an optional payment method in Stripe's dashboard.
- Tempo uses sessions to batch millions of agent transactions locally before settling on-chain, amortizing blockchain costs instead of confirming each payment individually.
- Stripe's existing merchant base bootstraps network effects, letting any merchant enable MPP with a dashboard toggle while Tempo targets fragmented developer APIs that OpenAI and Anthropic won't capture directly.
Summary
Paradigm and Stripe launched Tempo mainnet, a payments-focused blockchain designed to let AI agents transact autonomously. Georgios Konstantopoulos, who leads engineering at Tempo, identified the core problem: agents excel at searching and reasoning across the web but struggle to make payments. The solution is the Machine Payments Protocol (MPP), co-authored with Stripe and already available as an optional payment method in Stripe's documentation for any merchant to enable.
How it works
MPP complements the Model Context Protocol (MCP), which lets agents talk to APIs. MPP adds the ability to payment-gate those APIs. Tempo itself is optimized for speed with sub-second block times, but the real innovation for agent-scale payments is a feature called sessions. Rather than confirm each transaction on-chain, which would require millions of blockchain confirmations per second for agentic workloads, sessions let agents open a tab, rack up API calls locally, and settle only at the end. This amortizes the cost of on-chain settlement across hundreds of thousands or millions of calls.
Konstatopoulos also highlighted access keys, which function as programmable spend limits. An agent can be authorized to spend up to $10 per session, $1 per second, or any other granular constraint the user sets.
Use cases
Two categories emerged. First, commerce and content paywalls. An agent browsing ChatGPT could buy articles that would otherwise be paywalled, potentially improving research quality by 30%. Publishers could enable MPP directly through Stripe with minimal friction.
Second, and more important to Tempo's strategy, is agent-to-API payments between developers. Today accessing APIs requires logging in, adding billing info, and pulling API keys. With MPP, a developer can tell an agent "here's $5, go figure it out," and the agent orchestrates across multiple services—search the web, translate with one API, text-to-speech with another, upload to a website, send an email—all in a single prompt.
Distribution and network effects
The long tail of developers building paid APIs—knowledge bases, specialized tools—will be too fragmented for labs like OpenAI or Anthropic to capture directly. The lowest-friction path is MPP on Tempo. Stripe's existing merchant base provides a bootstrap network. Any Stripe merchant can flip a switch in the dashboard to enable MPP on their website or API.
Tempo offers a simple onboarding flow. Users install a Tempo skill in an agent like Claude, which prompts them to authenticate via Face ID or Touch ID and fund the wallet through Apple Pay, an Apple Card, or stablecoins. Users can then tell agents to add more funds on demand if they hit a paywall or need a higher budget.
The thesis
Tempo's bet rests on three premises. First, agents will make far more API calls than humans ever did, justifying a blockchain built for payment throughput rather than generalized compute. Second, Stripe's distribution advantage—merchants already on the platform—can bootstrap a payment network faster than building one from scratch. Third, as LLM models commoditize and traffic disperses across multiple sources, agents will need a programmable, censorship-resistant payment rail that doesn't require bilateral deals between labs and publishers.