Patrick Collison and John Collison, founders of the payments company Stripe, stated on social network X that blockchain networks will need to handle up to 1 billion transactions per second (TPS) to meet future demand from AI agents. Otherwise, current infrastructure will fail to keep up with the surge in automated online transactions.
The Collison brothers argue that AI agents will soon become the primary participants in most digital transactions, creating unprecedented stress on blockchain systems.
The Blockchain Infrastructure Gap
Stripe’s letter outlines a recent example: during a memecoin trading frenzy on a major network last year, payments were delayed for a Bridge user by over 12 hours, and transaction costs jumped 35-fold. This highlights the fragility of current systems when faced with sudden spikes in activity.
For perspective, top-performing networks today are Solana and Internet Computer, reaching approximately 1,140-1,196 TPS at peak. However, even their theoretical maximums of 65,000-209,708 TPS fall far short of the billion TPS needed for AI-scale operations.
Five Levels of AI Agents
The Stripe founders also outlined five levels of AI agent capabilities:
- Basic automation – filling out web forms and performing standard online tasks.
- Descriptive search – finding results based on situational descriptions rather than specific parameters.
- Context preservation – remembering user preferences, information, and requirements.
- Delegation – performing tasks on behalf of users, like online shopping.
- Anticipation – suggesting solutions or schedules without explicit prompts.
Currently, AI agents are hovering between levels one and two, with the Collisons noting that future progress depends on universal interoperability between systems.
The Path Forward for Blockchain
Stripe emphasizes that meeting AI demands will likely require a horizontal architecture of multiple, interacting chains rather than single high-TPS networks. Energy efficiency is another concern, as scaling to billions of transactions per second will require careful balancing of decentralization and operational costs.
As AI adoption accelerates, the first blockchain networks to achieve these scalability benchmarks could dominate the emerging market for automated digital transactions.
