Google restructures AI coding team as 3 top researchers depart

Google's restructuring of its AI coding team marks the company's most aggressive push yet to reclaim ground in a market where Anthropic and OpenAI have established commanding leads.

Google restructured its AI coding task force on June 25, intensifying efforts to capture a larger share of the profitable AI software market where Anthropic and OpenAI have pulled ahead.

"We have lots of things in the pipeline that will pay off short and long term," Logan Kilpatrick, a member of technical staff working on Gemini, said.

The reorganization follows the departure of three high-profile researchers. Noam Shazeer, co-inventor of the Transformer architecture that underpins most large language models, left for OpenAI. John Jumper, a Nobel laureate for his work on AlphaFold, joined Anthropic. Andrej Karpathy, an OpenAI co-founder who coined the term "vibe coding," also moved to Anthropic. Google's stock fell about 5% on the first trading day after the exits were announced.

The talent losses threaten Google's position in AI coding, which has emerged as the first major enterprise use case for generative AI. Google paid $2.7 billion in 2024 to license Character.AI's technology and rehire Shazeer, a figure that illustrates how much the company values the talent it is now losing to rivals.

Anthropic and OpenAI have established leads in coding benchmarks, a market that analysts project will generate billions in annual revenue as enterprises adopt AI-assisted development tools. Anthropic's Claude code assistant and OpenAI's Codex have become go-to tools for developers, while Google's Gemini has struggled to match their performance on key coding evaluations such as HumanEval and SWE-bench. The coding segment is particularly lucrative because it directly reduces enterprise engineering costs, making it one of the fastest-adopted AI use cases in corporate budgets.

Google's restructuring aims to consolidate its AI coding efforts — spread across DeepMind, Google Brain, and the broader organization — into a more focused unit capable of faster iteration. The company is betting that a unified team can move more quickly than the fragmented structure that allowed Anthropic and OpenAI to pull ahead. The new task force will prioritize developer tooling and code generation, areas where enterprise customers are already spending heavily on AI subscriptions.

The broader AI talent war has intensified across the industry. Last year, Meta recruited over a dozen top researchers from DeepMind and Scale AI with compensation packages that set new benchmarks for the field. The competition for celebrity researchers has escalated as labs recognize that a handful of star engineers can determine which company leads in critical AI capabilities. Anthropic, in particular, has been aggressive in poaching from Google, securing both Jumper and Karpathy within weeks of each other.

The company maintains a deep bench of talent, with top-level PhDs, researchers, and engineers across its AI divisions. "The talent density at Google is extremely high," Samira Khan, a Google DeepMind employee, said. "That leads to incredible internal competition for resources but also means there is inherent resiliency built into the system." Google also holds a significant stake in Anthropic, partially hedging any competitive gap between the two companies.

For investors, the question is whether Google's restructuring can close the coding gap before Anthropic and OpenAI solidify their leads. If Google's revamped coding team delivers competitive products, it could protect the company's enterprise AI revenue stream. If it falls further behind, the market may begin pricing in a permanent loss of developer mindshare — a risk that Alphabet's roughly $2 trillion market capitalization makes material for institutional investors.

This article is for informational purposes only and does not constitute investment advice.