Google Gemini Co-Lead Noam Shazeer Leaves for OpenAI After Billion-Dollar Rehire

According to reports Noam Shazeer, a leader in the AI field,  has left Google and accepted a new position at OpenAI. This move marks a significant development in the competition between AI talent as it represents yet another major loss of top researchers from top-tier AI companies. As one of the original architects of LLMs and cutting-edge transformer based architecture, Shazeer returned to Google after a high-profile separation that generated tens of billions of dollars in compensation.

In less than two years since his last return to Google, the competition among AI research Labs for top talent is intensifying due to an increasing level of external competition for high-performing AI researchers.

Why is Noam Shazeer Moving to Open AI?

It has been reported that Sam Altman, who is the Chief Executive Officer at OpenAI, has pursued acquiring the services of Noam Shazeer as he, along with being one of the leaders in AI research today is also one of the most talented researchers in the industry. Reports indicate that Altman has been trying to recruit Shazeer for a prolonged period of time, indicating how serious OpenAI was about acquiring top talent in an area like Artificial Intelligence where there continues to be an increasing level of competition.

Shazeer has made critical contributions to the foundational architecture behind the all-important transformer models used to develop large-scale language models such as BERT and GPT-3 and he has been a major contributor to many of the generative AI capabilities that we see with today's generative AI solutions (including many of the capabilities tied to Gemini). His expertise in optimising the scaling and performance of large-scale models (e.g., transformer architecture) has made Shazeer one of the most highly sought after and best compensated researchers on a global basis.

Noam Shazeer’s Role in Google Gemini and AI Research 

One of the leading research figures in the field of modern artificial intelligence research, Noam Shazeer has been instrumental in shaping the direction of Gemini at Google. He is probably best known for co-authoring the groundbreaking "Attention Is All You Need" paper, which introduced the transformer model architecture that powers nearly all of today's large-scale language models.

Within Google's AI organization, he has been an integral member of the design team of large-scale AI model archtiectures with a focus on efficiency, scalability and performance. Shazeer has also been a major contributor to many techniques, such as mixture-of-experts models, which enable an AI system to activate only certain parts of its neural net(s) when making predictions thereby reducing the computational costs associated with making predictions.

Upon returning to Google as co-technical lead of Gemini, Shazeer assisted in developing next-generation AI systems capable of working with multiple data types. Additionally, his efforts have been crucial in ensuring that Gemini remains at the forefront of cutting-edge innovations and artificial intelligence, and industry leaders such as OpenAI.

Billion Dollar Rehire at Google Gemini

In 2000, Noam Shazeer began working at Google at the very start of its rapid growth and continued to work at Google for over 10 years as an AI researcher. He contributed to many of the core ideas behind the use of neural networks and was instrumental in the building of transformer-type network architecture leading up to the creation of large modern-day language models.

In 2021 he left Google and started Character AI, where he focuses on making consumer-friendly programs based on AI chatbots. After being away from Google for several years, he was rehired in a high-esteem position with reported large compensation in 2024.

When he was rehired at Google, he became a co-lead on the Gemini program working with others to continue developing large multimodal AI systems at a time when competition with OpenAI was fierce.