The race for leadership in artificial intelligence (AI) has led Big Tech companies to spend billions of dollars not only on infrastructure but also on talent, generating intense competition for the best professionals to develop language models.
The most recent development became public on Thursday, June 18, when Noam Shazeer announced he would be leaving Google to work at OpenAI .
“I’m very happy to share that I will be joining OpenAI and I look forward to working with the exceptional team there,” said the Google vice president of engineering and co-lead of Gemini AI models, in a post on X.
To date, there has been no announcement from OpenAI nor any information about the position he will hold at Sam Altman 's company.
Shazeer's departure represents a major blow to Google, which paid dearly to bring him back less than two years after his exit, demonstrating how intense the competition is for professionals with expertise in AI.
In 2024, Google wrote a check for $2.7 billion for the chatbot startup Character.AI, co-founded by Shazeer. This was the company's way of bringing its former employee back, along with his research team, to strengthen its AI division.
Big Tech companies are paying billions of dollars for AI startups to recruit specialists in the technology, which is still scarce in the market. To do this, they are resorting to different formats, not just the traditional headhunter with a high salary offer.
One of these strategies is the so-called acqui-hire , used by Google with Shazeer. This is a corporate strategy in which a company buys another primarily to absorb its talent, and not necessarily its products or services. It's the "M&A of people".
Big Tech companies have also resorted to a model in which they hire founders and key AI researchers from startups and license the technology from those companies.
Noam Shazeer helped put Google back in the AI race .
Microsoft did something similar in 2024 with Inflection AI, hiring its CEO, Mustafa Suleyman, to lead the Copilot AI unit. In addition, it paid the company a licensing fee of $650 million, according to information from The Wall Street Journal (WSJ) .
The talent shortage also causes professionals to frequently change companies in search of the best offer. In 2025, Apple lost Ruoming Pang, its top AI modeling executive, who moved to Meta.
A Bloomberg report indicated that Mark Zuckerberg 's company offered a package valued at tens of millions of dollars per year to attract Pang. The company has been investing heavily in AI — in 2023 alone, Meta allocated US$78.4 billion to the technology, an amount also used to recruit professionals from OpenAI and Anthropic .
Despite paying a high price for Pang, he didn't stay at Meta for very long. According to information from The Information website, he was hired by OpenAI at the beginning of the year, seven months after starting his career at Meta.
In the case of Shazeer, OpenAI is bringing in a professional considered one of the pioneers of AI. More than helping to put Google back in the AI race, reducing the gap with ChatGPT , he is one of those responsible for formulating the concepts that gave rise to generative AI.
In 2017, while still at Google, he co-authored a paper on Transformer architecture, a way of building AI systems capable of understanding and generating language (and other types of data) by simultaneously analyzing the relationships between different pieces of information.
The publication is considered one of the factors that enabled the boom in generative AI, by presenting an architecture capable of identifying which parts of a piece of information are most important for understanding the context, serving as the basis for all the most popular language models today.
