What Happened to Google Brain?

The Google Brain project began in 2011 as a research collaboration at Google X, involving Google researchers Jeff Dean and Greg Corrado, and Stanford professor Andrew Ng. Its initial goal was to build deep learning processes on top of Google's existing infrastructure. A notable early achievement in 2012 was training a large-scale neural network on YouTube videos, which learned to recognize high-level concepts like cats without explicit training, demonstrating the power of deep learning.

Over the years, Google Brain became a central player in AI innovation, creating open-source tools like TensorFlow in 2015, which democratized access to machine learning model development. In 2017, the team introduced the Transformer architecture, a groundbreaking neural network design that became the dominant architecture for large language models and underpins many modern AI systems, including OpenAI's GPT series. Google Brain also contributed to various Google products, including Google Translate, Android's speech recognition, and YouTube recommendations, and explored areas like robotics, medical applications, and text-to-image models such as Imagen.

In April 2023, in a significant company-wide reorganization, Google Brain merged with DeepMind, a British AI research subsidiary acquired by Google in 2014, to form Google DeepMind. This move was largely a response to the accelerating AI race, particularly after the public release of OpenAI's ChatGPT, aiming to consolidate resources and accelerate the development of next-generation general AI systems. Demis Hassabis, DeepMind's co-founder, became the CEO of the newly formed Google DeepMind.

Post-merger, Google DeepMind launched the Gemini family of large language models in December 2023, positioning it as a major competitor to other frontier AI models. The combined entity continued to release advanced models like Gemini 3 Pro (November 2025), Gemini 3.1 Pro (February 2026), and Gemma 4 (April 2026), alongside specialized AI systems like AlphaFold. However, the merger also presented structural challenges and cultural clashes between the product-driven focus of Google Brain and DeepMind's pure scientific discovery approach.

A major leadership reshuffle occurred on August 5, 2026. Jeff Dean, a co-founder of Google Brain and Google's chief scientist, departed after 27 years to co-found a new startup, Discovery Loop, taking other senior researchers with him. Concurrently, Demis Hassabis stepped back from daily operations of Google DeepMind to become its Chairman and Alphabet's Chief Scientist, with Koray Kavukcuoglu taking over as Senior Vice President, reporting directly to Sundar Pichai. This reorganization was interpreted as an attempt to prioritize execution and commercialization over academic research, effectively dissolving the two-continent split that had characterized Google AI since the 2023 merger. As of September 2026, Google DeepMind continues to push the frontier of AI, with researchers noting early signs of recursive self-improvement in AI development and releasing models like Gemini 3.8 Flash. The company also launched the DeepMind Institute in September 2026 to explore the societal impacts of AGI.