Perplexity CEO Aravind Srinivas introduces a "hybrid local server agentic inference orchestrator" developed with Intel to optimize AI workloads by dynamically routing tasks between local devices and cloud servers. This software functions as an operating system that balances privacy, accuracy, and cost, allowing users to leverage both local file systems and frontier models like those from OpenAI or Anthropic. Despite competition from these labs, Perplexity’s revenue has tripled since the start of 2024, surpassing a $500 million annual revenue run rate as power users increasingly adopt high-tier subscription plans for advanced research. Addressing legal challenges regarding copyright, Srinivas maintains that facts and truth are not subject to copyright ownership. The platform's success is measured by user retention and the ability to provide immediate, accurate answers rather than maximizing time spent on the interface.
Sign in to continue reading, translating and more.
Open full episode in Podwise