Ex-Googler Claims DeepMind Uses Claude While Rest of Google Does Not

A Public Dispute Erupts

A simmering row over AI tool access inside Google has burst into the open, pitting software engineering veteran Steve Yegge against some of the company's most senior leaders, including Google DeepMind CEO Demis Hassabis. The controversy began on 13 April when Yegge, a former Google engineer, posted on X that a longtime Google tech director had told him internal AI adoption at the company was unremarkable, comparable to that of 'John Deere, the tractor company.' He outlined what he called an industry-standard pattern: 20% of engineers as 'agentic power users,' 60% relying on basic chat tools, and 20% outright refusing to use AI at all.

Google Leadership Pushes Back

The post drew swift rebukes. Hassabis called Yegge's claims 'completely false and just pure clickbait' and advised Yegge's source to do 'actual work.' Addy Osmani, a director at Google Cloud, countered that more than 40,000 Google software engineers use agentic coding tools on a weekly basis, insisting 'Google is anything but average.'

The Two-Tier Allegation

Yegge escalated his claims on Monday, writing that he had heard from 'Googlers from multiple orgs' who described a 'two-tier system.' Under this arrangement, DeepMind engineers use Anthropic's Claude as a daily coding tool, while the rest of Google's engineering workforce is steered toward internal Gemini models. Yegge wrote that when the question of equalising access arose internally, the proposed solution was to remove Claude for everyone, a move DeepMind objected to. He dismissed the 40,000-engineer figure, arguing that 'weekly use of a thin tool is precisely the box-checking I described.'

Why This Matters

The dispute touches a raw nerve. Anthropic's Claude Code has emerged as a dominant force in AI-assisted software development, generating over $2.5 billion in annualised revenue by February. At the recent HumanX conference in San Francisco, Claude Code was reportedly the tool 'on everyone's lips.' Google has also reportedly assembled a dedicated 'strike team' of researchers and engineers to improve its AI coding models, reflecting growing urgency inside the company to keep pace with rivals, particularly Anthropic, whose tools have reshaped the competitive landscape. Google has not commented publicly on Yegge's specific claims about tiered access.