Vinton Cerf, one of the architects of the networking protocols underlying the modern internet, will step down as Google’s chief internet evangelist next week, marking the end of a decades-long career in technology.
The announcement came in 2026 during the Open Frontier conference hosted by the Laude Institute, where Cerf appeared via video feed. Dave Patterson, a UC Berkeley professor known for co-developing RISC processor architecture, made the retirement public during the event. “Vint has been at Google more than 20 years, and he is retiring a week from today, and so I think we ought to give him a round of applause for a relatively good career,” Patterson said, to applause from the room. Google did not respond to a request for comment.
Cerf, 83, and collaborator Robert Kahn are credited with developing TCP/IP — the set of rules that allows different computer networks to communicate — beginning in the 1970s. That work has earned Cerf numerous honorary degrees, a Presidential Medal of Freedom, and a Turing Award. He joined Google in 2005 as a vice president and chief internet evangelist.
Cerf was speaking on a panel alongside other computer scientists, including François Chollet, creator of the Keras deep-learning library and co-founder of Ndea; John Ousterhout, the Stanford computer scientist behind the Tcl programming language; and Matei Zaharia, co-founder and chief technologist of Databricks. Discussion at the conference centered on the centralization of advanced AI models versus the decentralized architecture that made early internet protocols resilient.
Cerf offered a pointed prediction about where AI is heading. He argued that the rise of AI agents — software that acts autonomously and coordinates with other software — would push the industry toward formal interoperability standards. “The agentic model of AI, with multiple agents from multiple sources interacting with each other, is going to force composability, and a requirement for interoperability and standardization,” he said.
Cerf also pushed back against the idea that natural language alone could handle communication between AI agents, warning that ambiguity in English could compound across agent interactions. “Imagine a bunch of agents talking to each other in natural language — that’s kind of terrifying,” he said.
Source: TechCrunch