Two leading artificial intelligence researchers at Alphabet Inc.’s Google are planning to leave for rival Anthropic PBC, according to people familiar with the matter, adding to a series of high-profile departures that risk undermining the search giant’s position in artificial intelligence.
Jonas Adler and Alexander Pritzell, who are considered by the company to be key contributors to Google’s Gemini AI model, will be joining Claude developers, the people said, asking not to be identified because the information is private.
Mr. Adler worked on the company’s AI coding efforts, and Mr. Pritzell was involved in the process of training the artificial intelligence system.
Google, an early pioneer in artificial intelligence, spent much of the current AI boom catching up with companies like OpenAI and Anthropic before coming into its own late last year with more capable models and chips. But the company has already lost two notable staffers in recent days, with Nobel Prize winner John Jumper moving to Anthropic and star researcher Noam Shazeer moving to OpenAI.
Their moves spooked investors and cast new doubts on Google’s ability to participate in the fierce race to build better models.
Alphabet shares closed slightly lower after falling as much as 1.2% during the trading day on Wednesday. The exit underscores the pressure Google is under from two startups about to go public, giving even wealthy employees at big tech companies a chance at a rare payday with pre-IPO deals. In at least one case, Google’s departure appears to have been preceded by a change in priorities about how to allocate valuable computing resources, an issue that led to other employees leaving the company entirely.
Shortly before Shazeer announced his plans to join OpenAI, computing power devoted to one of his projects was reallocated to Google’s London-based team at DeepMind, according to two people familiar with the matter. The move was made to foster collaboration between teams and streamline Google’s work in pre-training, the early stage of AI development where models learn from large datasets, the people said.
Mr. Adler, Mr. Pritzell, Mr. Jumper and Mr. Shazier did not respond to requests for comment. Antropic declined to comment. A Google spokesperson said the company remains confident in its position in the AI talent market, pointing to comments from Google DeepMind CEO Demis Hassabis earlier this week.
“There is a high flow of talent among leading labs, and we get our fair share of the best talent. We have by far the largest and most extensive research bench of any lab in the world,” Hassabis said at a Cannes event. “The current market is fiercely competitive, the most competitive the technology industry has ever seen.”
Shazeer’s career trajectory is emblematic of the fierce talent wars that have defined the world of AI.
After co-authoring a seminal paper that helped fuel the AI boom, he left Google in 2021 to found chatbot startup Character.AI, but returned to the company in 2024 as part of an unusual licensing deal that valued the company at $2.5 billion.
Back at Google, Shazeer co-led the development of the company’s flagship Gemini AI model. Two of the people said he was also working on a new AI architecture before leaving the company. This architecture is still based on Transformers. The transformers, a technology Shazeer and others introduced in 2017 and have since become a staple of AI development, have yielded promising results, one of the people said.
Mr. Shazier was both admired and polarizing within Google, according to current and former employees. His comments at Google about transgender identity and the conflict in Gaza sparked controversy among some employees, two people familiar with the matter said.
Meanwhile, Jumper had emerged as the face of Google’s most ambitious AI efforts after winning a Nobel Prize for his groundbreaking work using AI to predict protein folding. Both Adler and Pretzel will be joining Jumper on Anthropic and worked with him on its research.
Key members of Jumper’s team studying protein folding have left Google DeepMind in recent months. Some went to Isomorphic Labs, an Alphabet spinout that develops AI-designed drugs, according to people familiar with the matter.
Anthropic competes with and partners with Google, but it also actively siphons talent from the tech giant. DeepMind engineers are nearly 11 times more likely to leave for Anthropic than vice versa, according to a 2025 industry analysis from venture capital firm SignalFire.
Like Google, Claude manufacturers are also exploring life science and healthcare applications to expand the use of their technology. Anthropic recently raised new funding at a valuation of $965 billion, more than OpenAI, and is considering going public as early as this fall. AI researchers in the UK, where DeepMind’s leadership is based, are often subject to long-term non-compete agreements enforceable under UK law.
Mr. Jumper will likely not start working at Anthropic until next year, according to a person familiar with the matter. Another researcher, Arthur Conmee, wrote on Wednesday that he plans to join Anthropic to work on AI safety. According to his LinkedIn profile, Conmy was a senior research engineer while at DeepMind, where he contributed to Gemini 2.5 models and AI coding.
Love, Mascarenhas and Metz write for Bloomberg.
