As large language model (LLM)-based agents increasingly perform tasks on the internet, they are generating significant web traffic without contributing to ad or subscription revenue, posing a challenge to the economic sustainability of the open web. The study highlights that these AI agents can browse hundreds of pages per second, dramatically increasing web traffic (10-60 times more than humans for similar queries) while bypassing traditional revenue models, as they do not interact with advertisements or paywalls. This behavior, termed "agentic search," suggests that content creators and publishers are not compensated for the value extracted by AI systems, which are trained on open web data. The proliferation of such tools could lead to fragmentation of the open web unless new monetization models, like micropayments facilitated by systems like USDC, are developed to maintain its viability. The analysis underscores the need for sustainable economic models to support the integrity of web resources as AI-powered search capabilities continue to expand.