At London’s Breakpoint conference, Solana Foundation AI lead Rishin Sharma pointed to payments as an early use case for AI agents. As reported on November 17, 2026, he said the x402 standard for agent payments had handled more than 400 million payments totaling $100 million in transaction volume. He also said BlockRun had processed more than 60 million transactions on Solana over the preceding two months.
Sharma presented Solana.ai, a portal launched the previous week that brought together more than 120 teams building AI products on Solana across payments, trading and infrastructure. He said the foundation had also introduced payment channels described as capable of handling 1 million payments per second, an AI service offering private inference, and Pay.sh, which integrates more than 100 APIs. The report did not provide an independent throughput measurement for the payment channels.
To explain the foundation’s focus, Sharma cited figures indicating that automated software accounted for more than 53% of global web traffic in 2025. He said API keys classified as agents accounted for 76% of tokens processed by model marketplace OpenRouter in October 2026, up from 8% a year earlier. He also contrasted a quoted cost of $37.50 per million tokens for GPT-4 with $0.14 for GPT-5 nano, while saying monthly token use had grown roughly 330-fold.
The foundation had previously partnered with Google Cloud in May on a stablecoin payment service for AI agents. It also said it was working with Google, Alibaba and ByteDance; the report did not specify what those collaborations involved. Separately, a foundation backed by Coinbase and the Linux Foundation was advancing the x402 standard.
Raoul Pal, founder of Real Vision, offered a longer-term view at the same conference: he said AI agents were not yet genuine economic participants, but argued that agents would use blockchains for trading and settlement as assets such as stocks move on-chain. That was his projection, not a description of current agent activity.