Sergey Brin calls retirement 'the worst decision' and returns to Google to lead Gemini AI push

Google Co-Founder Admits Strategic Misstep on AI

Sergey Brin, the Google co-founder with a net worth of $247 billion, has publicly acknowledged what he calls a critical error in judgment—both personal and corporate. Speaking at Stanford University's School of Engineering centennial celebration, Brin revealed that his decision to retire from Google in 2019 was "the worst decision" and that the company "for sure messed up" by failing to capitalize on its early lead in AI 1

1

3

3

. The admission marks a rare moment of candor from a tech founder about missing a technological wave that Google itself helped create.

Brin had imagined retirement as quiet days studying physics in cafés, but the COVID-19 pandemic shut down that vision almost immediately. Without the intellectually demanding work he was accustomed to, he found himself "spiraling" and "not being sharp," describing the experience as "just kind of stewing" in isolation 2

2

. The transition from president of Alphabet to a 40-something retiree proved far more difficult than anticipated.

Return From Retirement Fueled by AI Arms Race

When Google began allowing employees back into offices during the pandemic, Brin seized the opportunity. He started spending time on what would become Gemini AI, Google's flagship large language models designed to compete in the AI arms race 1

1

. By 2023, Brin was visiting the office three to four times a week, working directly with researchers, holding weekly discussions about AI research, and even participating in personnel decisions like hiring. "To be able to have that technical creative outlet, I think that's very rewarding," Brin said, adding that staying retired "would've been a big mistake" 1

1

.

The urgency of his return from retirement stems from a strategic blunder Brin now openly acknowledges. Despite Google researchers publishing the groundbreaking Transformer paper eight years ago—the very architecture that became the "T" in ChatGPT—the company "didn't take it as seriously as we should have" 3

3

. Corporate caution played a role, with Google "too scared" to release chatbots publicly because early models had a tendency to say "dumb things." While Google kept its AI research confined to the lab, OpenAI moved aggressively to productize the technology, catching the tech giant off guard with ChatGPT's release in 2022 3

3

.

Google Messed Up Despite Pioneering Neural Network Research

The irony is stark: Google had led AI research for years through DeepMind and Google Brain, yet failed to anticipate that "Deep Tech" would become the industry's primary battlefield so quickly 3

3

. Brin suggested the company rested on its laurels and underinvested in the very technology it pioneered. Now fully back at work, Brin has pushed for an intense work culture. In a February memo, he recommended Google employees working on Gemini AI come to the office "at least every weekday" and suggested 60-hour workweeks as the "sweet spot of productivity," despite Google's official three-day in-office policy 1

1

2

2

.

Gemini AI Shows Promise as OpenAI Declares Code Red

The pivot appears to be yielding results. Analysts praised Gemini 3, Google's latest update, and the company's stock rose about 8% following its release 1

1

. Meanwhile, OpenAI declared a "code red"—its highest alert level—to improve ChatGPT, suggesting the competitive landscape is shifting . Brin maintains that Google has advantages in the AI arms race due to its foundation in neural network research, custom AI chips, and data center infrastructure built over years 1

1

.

Brin now talks to Gemini AI during his commute to brainstorm ideas, emphasizing that the company has fully pivoted to prioritize AI development 3

3

. The world's third-richest person, with a net worth of $246 billion according to Bloomberg Billionaires Index, also tied his retirement regret to a broader pattern of founders misjudging timing—pointing to the $1,500 Google Glass as an example of rushing a product to market before it was ready 2

2

.