Imagine discovering the world's leading tech company, the one racing to dominate artificial intelligence, might be struggling to use its own tools. That's the bombshell allegation from a former Google engineer that has ignited a firestorm, pitting veteran insiders against the company's top brass in a very public and bitter spat.
Steve Yegge, a respected software engineering veteran, didn't just criticise Google quietly. He took to social media with a claim so provocative it forced Google's AI CEO to personally hit back. The core accusation? That Google's internal adoption of AI is shockingly low, comparable to a traditional tractor company, and that a secret divide exists within its own walls.
CEO's Fury and the "Absolute Nonsense" Retort
The reaction was swift and fierce. Demis Hassabis, the CEO of Google's elite AI unit DeepMind, didn't mince words. In a rare public rebuttal, he blasted the claims as "completely false and just pure clickbait," telling Yegge to tell his source to "do some actual work." This level of executive pushback is almost unheard of, signalling just how sensitive this nerve has been struck.
The "Two-Tier System" Insiders Are Scared to Expose
But instead of backing down, Yegge doubled down. He says that in the week following his initial post, multiple current Google employees reached out to him—anonymously and fearful of being exposed. What they described, according to Yegge, is a deeply divided engineering culture.
"DeepMind engineers use Claude as a daily tool. Most of the rest of Google does not," Yegge wrote. He claims that while Hassabis's team uses the industry-favoured Anthropic Claude, the majority of other engineers are "pushed onto internal Gemini variants." This, he argues, is not the sign of a healthy, AI-fluent organisation.
40,000 Users or Just Box-Ticking? The Statistic Wars
Google's defence came from another high-ranking director. Addy Osmani of Google Cloud stated that over 40,000 Google software engineers use "agentic coding" weekly, calling Yegge's picture inaccurate. But Yegge fired back, arguing that "weekly use of a thin tool is precisely the box-checking I described." He dismissed the metric, saying it includes people who tried it once and reverted to old habits, and that true adoption is measured in deep, daily usage.
Why This Feud Exposes a Much Bigger Industry Problem
This isn't just corporate gossip. It cuts to the heart of the biggest challenge in Silicon Valley: getting employees to truly embrace and use AI tools daily. Companies like Meta have created leaderboards to track usage, turning adoption into a competition. If even Google, with all its resources, is facing this struggle internally, it reveals a massive gap between public AI hype and private, practical reality.
The fallout from this clash is far from over. It forces a critical question for the entire tech industry: are companies winning the AI race on stage, while their own engineers are stuck at the starting line? For now, Yegge has drawn a line in the sand: "You can choose to believe Google's AI PR team... or you can believe your friends who actually work there." The credibility of one of the world's most powerful companies may hinge on the answer.