Google and Pichai Bet on Giant Models to Power Gemini

Google and Pichai Bet on Giant Models to Propel Gemini

Google CEO Sundar Pichai claims that the next leap for Gemini depends on building much larger foundational models.

Alphabet has raised its investment forecasts for 2026 to between $195 billion and $205 billion, up from a previous range of $180 billion to $190 billion. The company states that demand continues to outpace investment.

Google reported a revenue of $119.8 billion for the second quarter of 2026, a 24% increase from the previous year, surpassing analyst expectations of $116.9 billion. Google Cloud experienced an 82% growth, reaching $24.8 billion, while the Services segment grew by 15% to $94.5 billion. The Gemini app now boasts 950 million monthly active users, up from 750 million in February.

Alphabet's AI metrics for Q2 2026 show that Google Cloud has grown by 82% and that the Gemini app has reached 950 million monthly active users.

The AI mode in Google Search has surpassed 1 billion monthly active users since its global launch in October. Google claims that the AI mode, like the AI summaries before it, generates more global search queries. This should not be confused with an increase in traffic to websites compared to traditional search.

Google states that it sends billions of clicks to websites each week through AI features in Search, but without specific data or comparisons to traditional search volume, this claim remains somewhat uninformative. On the advertising side, AI Max has exited beta and already has 500,000 advertisers using it. Alphabet indicates that AI is unlocking billions of new search queries that were previously difficult to monetize.

The cost per response in AI mode has reached its lowest level since its launch, even as Google employs more powerful models. This indicates increasing efficiency gains, as also shown by the recent Flash 3.6 release. Over time, this efficiency could give Google an edge in margins for its AI service.

This also raises the question of how important a cutting-edge model is for Google on a daily basis. For the race towards AGI, this is likely significant. For serving everyday use cases in its ecosystem, it is probably less so. "We are seeing strong demand for our Gemini Flash series, as it strikes the right balance between performance and cost," said Sundar Pichai.

Google Bets on Gemini 4 to Close the Gap with Leading Rivals

Google is still working on a new cutting-edge model and plans to launch new models every month. However, Gemini 3.5 Pro apparently does not fit into this schedule, even though Google states it is currently in testing.

CEO Sundar Pichai acknowledged during the earnings call that Google needs a more powerful foundational model to catch up with the leaders. The next generation of cutting-edge AI models requires much larger foundational models, he said. Google is training Gemini 4 to achieve this goal.

"We have started our most ambitious pre-training run to date for Gemini 4, and we are excited about the progress we are seeing at the frontier," Pichai stated.

He also highlighted specific weaknesses that Google is trying to address. "There are areas where we have acknowledged that we need to improve. Coding and agentic coding are one example, and I think the teams are very, very focused on that."