Google DeepMind CEO Rejects Ex-Engineer's Claims of Slow Internal AI Adoption

Google DeepMind CEO Demis Hassabis publicly rejected claims by former Google engineer Steve Yegge that the company's internal AI adoption mirrors that of tractor maker John Deere. In a reply on X on April 14, 2026, Hassabis stated, 'Maybe tell your buddy to do some actual work and to stop spreading absolute nonsense.

' Yegge, a software engineering veteran, based his initial assertion on information from a buddy at Google who has been a tech director there for about 20 years. Yegge wrote on X on April 13, 2026, that Google's internal AI adoption curve is the same as John Deere's.

He added that most of the industry has an internal AI adoption curve of 20% agentic power users, 20% outright refusers, and 60% still using Cursor or equivalent chat tool. Business Insider reported it has not independently verified Yegge's claims, which he presented as secondhand, and Google did not respond to a request for comment from Business Insider.

On April 20, 2026, Yegge wrote on X that he had heard from Googlers from multiple orgs who outlined a two-tier system in which DeepMind engineers use Anthropic's Claude frequently and the rest of Google largely does not. He stated that non-DeepMind engineers get pushed onto internal Gemini variants.

In response to Hassabis earlier, Yegge said he would apologize for his initial post if Google can convince him that half their engineers are burning 4M tokens a day.

Addy Osmani, director of Google Cloud, wrote on X on April 13, 2026, that over 40,000 Google software engineers use agentic coding weekly. Osmani stated that Yegge's original claims did not match the state of agentic coding at Google. He added that Googlers have access to their own versions of antigravity, geminicli, custom models, skills, CLIs, and MCPs.