Financial technology giant Stripe has issued a stark warning about the future scalability of blockchain networks, stating they may need to process up to 1 billion transactions per second (TPS) to support the impending wave of artificial intelligence (AI) agents conducting online commerce. The prediction was made in an annual letter posted to X by Stripe CEO and co-founder Patrick Collison and co-founder John Collison, who reviewed the firm's 2025 performance and outlined key technological forecasts.
The core argument is that blockchain transaction activity will "skyrocket" as AI agents gradually become the primary conductors of online transactions. The co-founders highlighted a significant infrastructure gap, noting that current blockchain systems are ill-equipped for this incoming demand. They cited a recent incident where a memecoin trading frenzy on a major blockchain delayed payouts for a Bridge user by over 12 hours and spiked per-transaction prices by 35 times. "While such operational issues are already significant, they will only intensify, for we expect the appetite for transactions to grow a great deal," they wrote.
Stripe's vision is that "agents will most likely soon be responsible for most internet transactions," necessitating blockchains that support more than one million — or even one billion — TPS. For context, data from Chainspect shows the current leaders in transaction speed are Internet Computer Protocol (ICP) and Solana (SOL), processing roughly 1,196 and 1,140 TPS respectively. Their theoretical maximums are 209,708 TPS and 65,000 TPS, still orders of magnitude below the proposed AI-scale requirement.
Alongside the scalability challenge, the Stripe executives outlined a framework of five levels of AI agent capabilities, ranging from basic automation (Level 1: completing web forms) to advanced anticipation (Level 5: providing unsolicited solutions). They noted that AI commerce is past the "hype phase" and is now in a period of real-world experimentation, with agents currently "hovering on the edge" of Levels 1 and 2. The future success of this "agentic commerce" is contingent on universal interoperability, allowing AI agents to operate across open protocols rather than walled gardens.
Stripe suggests meeting this demand will likely require a horizontal architecture of multiple, interacting chains rather than relying on a single high-TPS network, and will also necessitate careful balancing of energy efficiency, decentralization, and operational costs. The first blockchain networks to achieve these scalability benchmarks could dominate the emerging market for automated digital transactions.
