The 2017 "Attention Is All You Need" paper, written by eight Google researchers, laid the groundwork for modern LLMs. In a striking example of the innovator's dilemma, every author left Google within a few years to start or join other AI companies, representing a massive failure to retain pivotal talent at a critical juncture.

The departures and role changes of key AI figures like Demis Hassabis and Jeff Dean signal a loss of confidence in Google's ability to compete in the AGI race, despite its foundational research contributions. Commentary suggests this is an expected but significant shakeup.

In his departure statement, Google AI legend Jeff Dean said his new company must build infrastructure "differently than how things are built at Google right now." This is a strong indictment of Google's internal systems, suggesting they are too rigid and bureaucratic for cutting-edge AI research.

DeepMind's lack of a dedicated CEO or financial reporting unit reveals Google's core AI strategy: treat it as a cross-organizational layer to be "vended" into existing products like Search and Docs. While this promotes integration, it may also prevent the focused, singular drive needed to compete with standalone AI companies, explaining why they often feel behind despite their capabilities.

Google is falling behind in the AI race not just due to technical challenges, but organizational dysfunction. Internal teams like Cloud, DeepMind, and Android are building competing AI coding tools, while some engineers' cultural resistance to AI-written code creates a slow, fragmented development process.

High-profile departures from DeepMind, like Nobel laureate John Jumper, are not isolated events. They are linked to plummeting internal morale caused by competitors like ZAI's GLM 5.2 overtaking them on benchmarks and a four-month drought of a flagship model release.

Google's direction is pulled between two philosophies. CEO Demis Hassabis favors a long-term, "world models" path to AGI, while a faction reportedly led by Sergey Brin pushes to compete directly with OpenAI and Anthropic on immediate applications like AI coding agents. This internal tension manifests as a confusing product roadmap.

Despite immense resources, Google is in danger of falling out of the top tier of AI labs. Its models are described as "deeply psychologically screwed up," its internal scaffolding efforts are weak, and its corporate culture hinders progress. This is causing them to lose ground to more focused competitors like Anthropic and OpenAI in the race for recursive self-improvement.

An on-leave DeepMind employee posted a thinly veiled allegory about Google's AI struggles, describing "slumbering execs" and a "King of terrorists" who holds the organization hostage. This suggests internal politics and ineffective leadership, not a lack of talent, are behind the company's perceived lag.