Google’s gearing up to roll out Gemini 3 next year. Sundar Pichai mentioned it during the recent earnings call, saying the team’s moving fast but needs more time for big leaps.
This new model should outperform Gemini 2.5 Pro, get closer to what GPT-5 can do, and focus more on handling complex tasks across text, images, and video—like a smart assistant that actually gets things done.
Pichai was clear, though: don’t expect miracles right away. “We’re excited about how quickly we’re improving things,” he said, “but building truly better models takes longer now.” Google usually drops new Gemini 3 versions in December, so that’s likely when we’ll see it.
Gemini’s become a huge part of Google’s AI push. In Q3 2025, Alphabet broke $100 billion in quarterly revenue for the first time, and Pichai told investors, “We’re fully in the generative AI era.”
Right now, Gemini has over 650 million monthly users, and queries jumped three times from last quarter. Still, it’s behind ChatGPT, which says it has 800 million weekly active users.
Google Search is growing fast, especially with younger people, thanks to AI Overviews and AI Mode. AI Mode now works in 40 languages and has 75 million daily users. In the U.S., queries through AI Mode doubled year-over-year last quarter.
But not everyone’s happy. Some site owners say Google’s AI summaries copy or rewrite their content, sending less traffic their way. Studies show people are clicking less on regular search links.
Google Cloud is a big growth engine. Revenue from generative AI products grew over 200% year-over-year. New cloud customers are up nearly 34%, and the backlog hit $155 billion—up 46% from last quarter.
More than 70% of Google’s cloud customers are using its AI tools, like Gemini, Imagen for images, and Veo for video. Gemini Enterprise, launched last month, already has 700 companies and over 2 million users worldwide.
Google’s building its own full AI stack—chips, models, software—all in-house to boost profits and keep customers locked in.
Demand for AI hardware is skyrocketing. Anthropic reportedly wants up to a million Google TPUs. Amin Vahdat, who runs AI infrastructure, said every TPU they have is booked solid. “We need to scale faster,” he admitted. The next-gen Ironwood TPU is coming soon, and they’re also launching new cloud services with NVIDIA’s GB300 chips.
On the research side, Google’s working on Veo 3 for video and Genie 3, a “world model” for simulating environments. So far, Veo’s created over 230 million videos, and more than 13 million developers are using Google’s AI tools.
Meanwhile, Waymo’s expanding fast. They plan to launch in London and Tokyo by 2026, and add Dallas, Denver, Seattle, and Nashville in the U.S. They already hold full driverless permits for airport runs in San Francisco and San Jose, and testing continues to grow in New York. To reach more people, they’ve started “Waymo for Business” and “Waymo Teens.”
Read Also: Samsung Exynos 2600 Just Leaked Geekbench Records: World’s First 2nm Chip
About The Author
