John and Patrick Collison on Stripe's Growth, Agent Commerce, and the Future of Software

Episode Summary

Executive Summary: Patrick and John Collison discuss Stripe’s 34% growth, a trillion-plus in payments, and what their data suggests about AI’s real-world impact. They argue agentic commerce is emerging, will require much higher-throughput blockchains, and may reshape software from fixed, mass-produced products into bespoke, on-demand creation. Stripe’s cohort data suggests the AI economy is accelerating, especially in 2025–2026.

Main Topics: Stripe’s growth and real-economy signals (Priority: 5/5): The Collisons explain that Stripe’s business and its customer cohorts are growing strongly, which they view as a more grounded indicator of economic health than market noise or headlines. AI adoption vs executive skepticism (Priority: 5/5): They push back on survey data suggesting executives see no AI value, arguing that real usage and purchasing behavior show meaningful productivity and business gains even if executives don’t fully perceive them. Agentic commerce and payment infrastructure (Priority: 5/5): They argue agents will increasingly perform real economic tasks, but today’s web and payment rails are not built for them, creating demand for new infrastructure and higher-throughput rails. Stablecoins, blockchains, and Tempo (Priority: 4/5): The conversation connects stablecoins with AI through the need for fast, scalable blockchains. Stripe’s incubation of Tempo is framed as a response to congestion and throughput limits. Product-first thinking vs MBA-style TAM analysis (Priority: 4/5): John emphasizes that successful new products come from solving concrete pain points rather than top-down market sizing or generic strategic logic. Software changes from fixed product to bespoke creation (Priority: 5/5): Patrick argues AI shifts software economics away from once-built, infinitely monetized products toward on-demand, customized software created at the moment of use, like pizza. Stripe Press and brand-building (Priority: 2/5): A lighter segment notes Stripe Press reaching over 1.1 million books sold, reinforcing Stripe’s broader identity as a long-term, culture-shaping company.

Key Arguments: Stripe’s data is showing that the economy, especially AI-enabled businesses, is stronger than public debate suggests. The 2025 cohort of businesses on Stripe is larger and performs better per business than prior cohorts, indicating a possible phase transition in AI adoption. Executives may underestimate AI value because usage is embedded in workflows or layers of the stack, making benefits less visible. Agentic commerce will expand beyond simple agent purchases into a torrent of real transactions, but this requires payment and blockchain infrastructure built for extremely high throughput. Stablecoins and AI are linked because agents need fast, global, programmable settlement rails. Tempo is meant to address blockchain congestion and support low-latency, high-throughput payments. New products should be built from specific customer pain points, not abstract TAM narratives or spreadsheet-driven planning. AI will alter software economics by making custom, inference-driven software more important than mass-produced software.

Data Points: Stripe payment volume processed: More than $1 trillion last year - Patrick cited Stripe’s annual scale as evidence of strong underlying commerce activity. Stripe growth rate: 34% - Stripe grew last year, driven by growth in businesses using the platform. 2025 business cohort: Larger and better performing on a per-business basis than any prior cohort - Patrick described this as a notable phase transition in the Stripe ecosystem. Timing of AI-driven economic inflection: End of 2025 / beginning of 2026 - Patrick said this is when Stripe is beginning to see AI’s effects in purchasing behavior. Predicted singularity timing: 2026 Q1 - Patrick said it may later be viewed as the first quarter of the singularity. Stripe Press books sold: 1.1 million - The team celebrated Stripe Press reaching this milestone. Throughput requirement for future platforms: Millions to billions of transactions per second - Patrick argued agentic commerce will require blockchain/payment infrastructure at this scale.

Pivotal Quotes: "2026 Q1 may be looked back on as the first quarter of the singularity." — Patrick Collison: He used this to frame AI’s recent acceleration and the possibility that the current moment will later look like a turning point. "We think there'll be a torrent of [agentic commerce]." — Patrick Collison: He described the likely scale of AI-driven commerce and why infrastructure must adapt. "Software should be cooked right then and there at the moment of use." — Patrick Collison: He used the pizza analogy to explain a future where software becomes bespoke and dynamically generated.

Implications: The discussion suggests AI’s biggest near-term effect may be transactional: shopping, payments, and workflow automation. If true, infrastructure for agents, stablecoins, and blockchains will matter as much as model quality, and software businesses may shift toward personalized, on-demand creation.