Google is now winning the AI war.
There has been something of a hype campaign around the release of Gemini 3 but people were skeptical. We have seen this before from OpenAI where they constantly hint that a new model is revolutionary and then release it and it’s not. So when Google was doing the same thing, AI watchers were understandably skeptical.
One of the leaks was some of the benchmarking. As you can see, it looks like it blows everything else out of the water.
Now for many users (especially non-coders) these kinds of metrics don’t always translate into usability. The trillions of dollars of investment going into AI isn’t about coding, it’s about everything.
In any case, I now believe those numbers are real.
As readers know, I have been very bullish on AI from the beginning and using AI models, though rarely (hardly ever) for writing. I don’t think people come here to read what ChatGPT thinks (though I do encourage people to use it for equity analysis).
I have also been extensively using the latest paid models from everyone for a variety of tasks. Recently, I bought a new house and a killer app for AI is analyzing rooms, decorating, making images of those plans, anticipating problems and planning/executing/advising DIY home projects.
With the current models, you can use a lot of back-and-forth plus some trial and error to get some very impressive results. However none were able to have much success with spacial reasoning tasks. I just tried Gemini 3 for the first time and it utterly crushed the problems I’d encountered before. Moreover, it almost immediately came to solutions that I’d spent weeks thinking about and working with LLMs on.
I’m impressed.
Now I’m certainly not going to be the final word on AI usability and I’ve only had maybe 40 minutes to use it as it was just soft launched into the Google AI Studio. Try it yourself. It looks to me like a big incremental step along the lines of when OpenAI went multimodal.
What I do think is that this is very important for markets. There is always the angst that AI has hit a wall but this is a big improvement and fixes some of the problems. Now I’m sure that there are flaws but this release is a big deal and a big step forward at first blush. I ultimately think that’s much more important than Nvidia’s earnings this week for the overall “AI trade”.
In terms of Google stock itself, shares rose on Monday on the filing showing Berkshire Hathaway bought shares. It’s also up again today despite a brutal tape.
I would be very worried if I’m Sam Altman today.
The path for Google to win in AI is clearer than ever:
- It can outspend everyone
- They seem to have people who know what they’re doing (unlike META)
- They were early in the AI game, and probably had LLMs before anyone
- They’ve been working on their own chips for longer than anyone
- They have the ecosystem and installed userbase
- They have the data with Gmail, YouTube and other avenues
- They can build it right into Chrome and Google.com
- They’re executing
I find it hard to see how anyone can challenge that.
There are still big questions about canabalization of search and advertising but Google is also the expert in integrating commercial advertising into context, which is the way that OpenAI wants to monetize. I don’t see how they can compete with Google on that front. Shares have had an incredible run but they’re still only trading at 25-27 forward earnings for 2026.
That’s a much more compelling valuation than Nvidia.
This article was written by Adam Button at investinglive.com.