GOOGLE BLEEDING ITS BEST AI RESEARCHERS AS TWO MORE GEMINI VETERANS HEAD TO ANTHROPIC Google keeps losing the people it can least afford to lose. Jonas Adler and Alexander Pritzel are the latest. Both are viewed internally as critical contributors to the Gemini model family. Adler led Google’s AI coding efforts. Pritzel worked on the training infrastructure that shapes how AI systems learn. They are heading to Anthropic. This follows a pattern that has become hard to ignore. Nobel laureate John Jumper left DeepMind for Anthropic earlier this year. Noam Shazeer, one of the original Transformer co-authors, left for OpenAI. Andrej Karpathy, once the face of Tesla’s AI operation, recently joined Anthropic’s pre-training team. The list keeps growing. What is happening here is not random attrition. Talented researchers are making deliberate choices about where the most interesting work is happening, where they have the most autonomy, and where they believe the best outcomes will emerge. Right now, those choices are pointing away from Google at a striking rate. Google is not standing still. The company is investing heavily in Gemini, in DeepMind, and in the infrastructure to compete at the frontier. But infrastructure and investment cannot replace the institutional knowledge and research instincts that walk out the door with people like Adler and Pritzel. For Anthropic, every hire at this level is a direct strengthening of the teams building the next Claude models.