Google engineers have been working on AI in recent weeks. Now that the big announcement is in the rearview mirror, CEO Sundar Pichai says it’s time to catch up on some sleep.
“I think some people need sleep,” Pichai said on the “Google AI: Release Notes” podcast on Wednesday. He added that he and his team would “hopefully get some rest” if possible.
On November 18th, Google released its latest AI model, Gemini 3, and the company is now on track to reach a market capitalization of $4 trillion. The company’s stock has soared nearly 70% this year, including a 12% jump after Gemini 3’s launch.
Gemini 3 has been well received. Salesforce CEO Marc Benioff said in a post about X this week that it is making “insane” leaps in reasoning, speed, and multimodal capabilities. He added that he has “no intention of going back” to ChatGPT after spending just “two hours” on Gemini 3.
The announcement renews the conversation about Google potentially becoming the new front-runner in the AI race, after years of ceding the spot to ChatGPT maker OpenAI.
Pichai said Google has been quietly laying the foundations for a long-term AI strategy for years.
“In 2016, I wanted to make the entire company AI-first,” Pichai said.
Between the development of Google Brain in 2012, the acquisition of DeepMind in 2014, AlphaGo’s victory in the Chinese board game Go, and the announcement of the first tensor processing unit (the proprietary internal chip used to train Gemini), the stage was set for the tech giant’s adoption of AI.
“When we saw this in 2016, it was clear that we were about to experience a new platform shift. It was a full-fledged bet to make Google an AI-first company,” he said.
But Pichai said the rapid adoption of generative AI presented an even bigger opportunity for the company, which was the start of Gemini. The company brought together its Google Brain and DeepMind teams, ramped up its AI infrastructure, and began moving even faster, he said.
Pichai said the core idea is to take a “full stack” approach to innovation by improving everything from infrastructure to model improvements in pre-training, post-training and test-time compute.
But approaching innovation takes time, Pichai said. When Google first tried to embrace the age of generative AI, it lacked capabilities and needed to invest in several areas to “scale up,” he said.
“From the outside looking in, it would look like we were silent or on the back foot, but we had all the foundational pieces in place and were running on top of that,” he said.
Since then, the tide has turned.
“We’re on the other side now,” he said.
