By 2023, many people, including the person writing this piece, wondered where Google AI was in the world? This was because OpenAI was eating Google’s lunch with its ChatGPT and was on its way to making the search giant irrelevant. Two years later in 2025, many people, including the person writing this piece, wonder if there is a company that can overtake Google in the AI ​​race. Gemini 3 and Nano Banana Pro, the two AI tools launched by Google this week, show that it will be difficult for OpenAI or any other AI company to defeat Google. ChatGPT 3, when it came on the scene, was a thorn in the side of Google that woke up a giant. Now, he is doing googly things that only he can do.

Gemini 3, Google’s new big language model, tops almost all AI benchmarks. It goes beyond ChatGPT 5 and Anthropic Cloud AI models. It has shown impressive results in benchmark after benchmark that can have clear real-world implications.

Meanwhile, the Nano Banana Pro aka Gemini 3 Pro image is nothing short of magic. In single signals it is creating infographics, designs and photos that would otherwise take a designer or illustrator several days. More importantly, it seems that the spelling and text errors made by the AI ​​image generator have been almost fixed. I mean people are using it to create fake IDs using its API (Nano Banana Pro in Gemini 3 does not allow this).

The two AI tools are better than anything OpenAI or any other AI company is offering right now. And I suspect an update to Veo 3, Google’s AI video generator, is next. It is also expected to exceed expectations like the Gemini 3 and Nano Banana Pro.

This raises the question, what will the AI ​​landscape look like in the coming months and years? This is all still open. A lot can happen in the world of AI within months or weeks. Or as the joke goes on the Internet: One day we’re back and the next day it’s gone. But whichever path the game takes, it’s clear that Google currently – and for the first time ever – has the lead in the AI ​​world. It’s ahead of OpenAI, so much so that even Sam Altman is worried. Or as he said in his note to OpenAI employees, there are going to be “temporary economic headwinds” and a “harsh environment” against OpenAI in the coming months.

The main problem for Google at first in AI was execution. Otherwise the company was always ahead of others. Its AI research team has been ahead of others to a great extent. In a way, the entire AI progress over the last 5 years is built on the research paper – Attention Is All You Need – that its team wrote. It has more data than others. It has, arguably, more compute power than others due to its own TPU. In fact, one of the major revelations of Gemini 3 is that the model was trained on a Google TPU and did not use any Nvidia GPUs. In other words, while the rest of the world chases after Nvidia GPUs and fights for them, Google clearly has its own AI processors that it can produce just as fast as TSMC can make them.

Now that Google has fixed the implementation, it is leaving others behind. The AI ​​race has become a game that Google has to lose.

So, can it still lose? Absolutely. There are unknown factors and AI remains a field that is rapidly evolving. There are two aspects in particular that deserve mention. There is a presence of a Chinese AI model. Compared to ChatGPT 5 and Gemini 3, AI systems like DeepSeek R1 and KMK2 are not the best. But they are surprisingly capable and close. More importantly, they are open-source models, which means their cost is negligible compared to ChatGPT, Cloud, and Gemini.

There is some politics going on here as China wants to disrupt the AI ​​sector for American companies. I’ll leave that part for another day. Right now, an important thing to note is that even if the Chinese AI models cannot defeat Gemini 3 and Google, they may still make it impossible for the company to win completely. In other words, it’s possible that Google, despite its edge in AI, won’t be able to dominate search the way it has.

And two, Grok and whatever meta is working. Grok is a dark horse, currently flying under the radar. This is a powerful AI model that can get much better in the future. Elon Musk and his companies, like Google, have abundant amounts of data as well as financial muscle and are focused on creating the kind of computation that can serve Grok well.

Meanwhile, Meta is working on something and yet we haven’t seen exactly what it is. All we know is that Mark Zuckerberg has spent billions of dollars building a superintelligence team. The outcome of that exercise is still awaited, and while there’s always the possibility that it won’t work for META, there’s a saying in Silicon Valley – never bet against Mark Zuckerberg.

No matter how the future AI wars play out, Google will be at the forefront. The company, for some surprising reasons, had taken it easy on AI between 2015 and 2022. But that phase is over now. A few years ago, when OpenAI and Microsoft joined hands to launch AI-powered Bing, Microsoft CEO Satya Nadella claimed to make Google dance. “I want people to know that we made Google dance,” Nadella quipped. This week reveals that Google the Giant is actually dancing now. This is beautiful to see.