Ex-Stripe executives raise $19.2M for AI agent payment protocol ATXP

Two former Stripe executives have raised $19.2 million to develop ATXP, a protocol designed to enable AI agents to make autonomous payments across the internet. Louis Amira, Stripe’s former head of crypto & AI partnerships, and David Noël-Romas, the company’s former head of crypto engineering, launched Circuit & Chisel to address what they see as a critical infrastructure gap preventing widespread AI agent adoption.

The big picture: ATXP aims to become the foundational protocol for agentic payments, similar to how HTTP enabled modern web browsing, potentially unlocking a new era of microtransactions powered by AI agents.

Why this matters: The protocol could enable AI agents to autonomously purchase information and services with tiny payments, fundamentally changing how the web operates and creating new revenue models for content providers.

• Agents could pay small amounts to access paywalled content, scrape private profiles, or gather real-time data from various sources.
• This infrastructure could trigger a surge of new AI applications, similar to how the iPhone’s launch sparked the iOS app ecosystem.

Key technical details: ATXP positions itself as a more neutral alternative to existing solutions like Coinbase’s x402 protocol, with backing from both competitors and collaborators in the space.

• The protocol will support Stripe’s stablecoin blockchain Tempo, integrating with the company’s broader crypto payments strategy.
• Circuit & Chisel plans to develop its own AI agents once the foundational protocol is established.

Funding landscape: The $19.2 million round attracted investors from Primary Venture Partners, ParaFi, and notably Coinbase Ventures, despite Coinbase developing competing technology.

• This cross-investment illustrates how companies that would typically compete are collaborating on foundational infrastructure.
• The funding comes as Google released its own open-source protocol last week in partnership with Coinbase for AI payment applications.

The challenge ahead: Amira acknowledges the “chicken and egg” problem of needing both the technical infrastructure and widespread adoption of microtransactions across the web.

What they’re saying: “The models are getting better, their brains are getting smarter, but they don’t have the right tools,” Amira explained, emphasizing that current AI capabilities are limited by infrastructure rather than intelligence.