Google CEO Pichai admits AI coding lag, warns AGI is approaching
TL;DR
In a Hard Fork podcast interview summarized by AIbase on May 25, 2026, Google CEO Sundar Pichai said Google is ahead in general intelligence and multimodality but "a bit behind" in agentic coding and long-horizon task handling compared to rivals. He outlined a gradual AI transformation of Search, continued TPU access for third parties, and noted that rapid progress means the expected arrival of AGI is "closer" than he previously thought.
About this summary
This article aggregates reporting from 3 news sources. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
Race to AGI Analysis
It’s unusual to hear a sitting Google CEO publicly admit the company is behind on anything, but Pichai is now on record saying Google lags rivals in “agentic coding” and complex long‑horizon tasks. That’s a polite way of acknowledging that tools like Cursor+Anthropic and OpenAI‑based coding agents have outrun Gemini in the workflows that matter to developers. At the same time, he doubles down on Google’s strengths: multimodal intelligence, scale, and a Search franchise that he insists will be transformed gradually rather than ripped up overnight.([aibase.com](https://www.aibase.com/zh/news/28308))
For the race to AGI, Pichai’s comments matter because they confirm two things insiders already suspected. First, Google is internally “AGI‑pilled”: he explicitly says timelines feel shorter than expected and warns society not to assume we have decades to prepare. Second, the company is trying to straddle two conflicting imperatives—protect a cash‑cow search business while racing to match or beat frontier labs on capabilities. That tension will shape how aggressively Google can push truly disruptive agents and how willing it is to cannibalize its own ad model in order to stay in the game.
