This was always the thing OpenAI needed to fear.
On Tuesday, Google announced Gemini 3, its latest AI model. Google says it’s better at coding, and much more creative — but that’s not what its rivals should be most worried about.
Google’s all-out offensive is putting the “Pro” version of Gemini 3 directly into the hands of users and developers, and, most significantly, it marks the first time Google is introducing its new AI model to search on day one. That means users can access Gemini 3 in Google search by clicking “AI mode,” rather than downloading an app or visiting a separate web page.
While OpenAI owns ChatGPT, the most popular AI chatbot on the planet, it doesn’t offer much else: it relies on partnerships for chips and data center space. Google, on the other hand, has it all, and can completely control its AI product from early-stage research to the in-house cloud that runs it to the YouTube recommendations and AI-generated search summaries that it delivers to users.
However, OpenAI still holds one major advantage over Google, and, in a twist of fate, it’s a branding game from which Google has benefited for years.
Google may have a full stack, but OpenAI has one big branding advantage
Google’s model is what it calls its “full-stack” advantage. It doesn’t have to lean on competitors the way many of its closest rivals do, nor get into the twisty game of circular financing.
Speaking on a roundtable ahead of Gemini 3’s launch, Google DeepMind CTO Koray Kavukcuoglu directly cited its end-to-end advantage as a reason for moving faster.
“One of the most important things for us at Google is this is possible because we have a very differentiated full-stack approach,” he said.
This is Google’s AI funnel right now:
- Google DeepMind researchers create the models.
- Google uses its in-house TPU chips to train those models.
- Google uses its own cloud service to host the models and make them available to developers.
- Google adds those models into search, YouTube, Gemini, and other popular apps, putting them directly into customers’ hands.
Google’s biggest challenge since late 2022 has been to get all of this in harmony, which was no easy task for an incumbent with almost 200,000 employees and a search advertising business it has spent decades trying not to disrupt.
The alignment has demanded structural changes, bigger investments in its cloud business, and an internal push to get leaner and faster.
It appears to finally be paying off.
Even with Google now operating at full throttle, it has a brand problem. The word “ChatGPT” has become synonymous with AI for many people, in the way “Google” became shorthand for searching the internet or Kleenex for facial tissue.
Lucky for Google, it has a lot of time to try to change that. While OpenAI had the first-mover advantage, Google has the cash-in-the-bank advantage. It already offers some of its models for free, and if Google wants to undercut rivals on price to get more of them into people’s hands, it could.
“There’s no debate that Google has all the technical intangibles across the stack (infrastructure, model, applications) and market channels to win,” wrote Bernstein analyst Mark Schmulik in a note last month, “but we need HARD evidence that they’re putting this all together.”
With Gemini 3, we may have just gotten some.
Have something to share? Contact this reporter via email at [email protected] or Signal at 628-228-1836. Use a personal email address and a non-work device; here’s our guide to sharing information securely.