4 min readNew DelhiJun 19, 2026 10:12 AM IST
First published on: Jun 18, 2026 at 07:58 AM IST
IPO-bound OpenAI seems to have secured yet another big name in its talent roster. Noam Shazeer, who was the vice president of engineering at Google and co-lead of the Gemini models, will be joining the Sam Altman-led AI company. Shazeer’s recruitment by ChatGPT-maker comes at a time when big tech is competing to obtain the best talent and introduce the most advanced models.
The AI researcher announced his exit through a post on X. “I’m excited to share that I’ll be joining OpenAI and look forward to working with the exceptional team there,” he wrote.
Shazeer is reportedly leaving in less than two years after Google paid about $2.7 billion to bring him on board. Earlier he was serving as the head of the startup – Character.AI, and had brought along a team of his researchers to Google.
Talking about his move in his post, Shazeer said that it was a difficult decision to move on adding that he is incredibly proud of the amazing team at Google and everything they built together. “ It has been an honor and a pleasure to work with all of you,” he wrote.
https://platform.x.com/widgets.js
Following his exit, in a statement to Reuters, the Search giant expressed that the company was grateful for Noam’s meaningful contributions to Google over the years.
OpenAI has steadily recruited some of Google DeepMind’s most respected researchers. In late 2024, Lucas Beyer, Alexander Kolesnikov and Xiaohua Zhai left Google DeepMind to join OpenAI, bringing expertise in computer vision, multimodal AI and large-scale foundation models.
These Google Deepmind alumni played important roles in advancing Vision Transformers (ViT) and other cutting-edge AI systems that power the modern image and multimodal models. In December 2025, OpenAI also hired Albert Lee, a veteran Google executive who steered corporate development efforts across Google Cloud and DeepMind.
In recent times, perhaps OpenAI’s biggest talent acquisition is likely Shazeer. He is one of the original authors of the 2017 reasearch paper on transformer architecture – Attention Is All You Need, helped lead Gemini at Google, founded Character.AI, and was re-hired by Google before now moving to OpenAI.
OpenAI, Google DeepMind, Anthropic and Meta are increasingly competing on talent rather than just models or compute prowess. Reuters reported that top AI researchers are now commanding compensation packages worth millions of dollars annually, with Google and OpenAI aggressively trying to recruit and retain these elite researchers.
Why is the talent-movement happening?
The AI industry is currently experiencing a trend of significant movement of top-level talent, primarily because elite researchers are regarded as essential for gaining access to computing infrastructure and laying the foundation of proprietary models.
Back in 2024, Google, Microsoft, and Amazon struck deals with AI startups, including Character. AI, Inflection, and Adept, allowing them to bring onboard their top researchers and executives without entirely acquiring the companies. The pool of researchers capable of frontier AI systems remains relatively small, rendering talent as one of the most key assets for these companies.
There could be numerous factors that are fuelling the high-profile movement of talents between AI companies. Large technology firms and AI labs have access to vast computing resources, proprietary data, massive funding and engineering talent, areas where startups fall short. At the same time, the rising cost of training and deploying advanced AI models has put strain on smaller firms to reach out to bigger firms for partnerships or align themselves with larger players.
On the other hand, for researchers, moving between leading AI companies can offer an opportunity to work on mega projects, gain access to larger-scale infrastructure and potentially influence the next generation of AI systems.
OpenAI’s latest move seems to be aimed at strengthening its research team as AI companies compete for leading talent and advanced models. (Express Image) 
Share
