Is ChatGPT Losing the Lead to Google – in the Final Stretch?
Something new is happening in the AI world right now. For years, we’ve talked about performance and “whose model is the smartest.” And sure, that’s interesting — but in practice it’s nearly impossible to measure objectively, and the debate always ends up drifting into subjective opinions.
In the end, the entire AI race comes down to one thing: whose product people will actually use.
And that’s a much more concrete question. Who is best positioned to get their solution distributed — to put it right where people already are, in their everyday routines?
Who really owns the distribution?
When ChatGPT exploded onto the scene, it wasn’t just because the model was impressive. The truly brilliant part was the interface — that simple little box where you could type anything. Calling it a “chat” feels obvious today, but at the time it was a new way of talking to a machine. It felt almost… human.
But that interface is starting to feel a bit old. A bit too much prompting. A bit too much adapting to the model’s way of working. AI is looking for new places to live: in browsers. In operating systems. In apps, cars, homes, devices.
“Well well — isn’t that Google waving over there?”
It is indeed. Because yes — Google already owns all of those touchpoints. Chrome, Android, in-car systems, home devices… And it’s telling that OpenAI and others seem to have reached the same conclusion. Like Perplexity before them, OpenAI recently launched its own AI browser (Atlas) in order to distribute their LLMs. But here’s the irony: both Perplexity Comet and OpenAI Atlas are built on Google’s Chromium. In practice, they’re simply variations of Google’s own browser, Chrome.
Which raises the question:
What happens when AI is no longer a destination, but an infrastructure? When it’s no longer something we go looking for, but a layer we simply live inside?
In the end — make what you want of it yourself. But to us, it’s hard not to sense a bit of desperation in the air. The initial magic of ChatGPT may be fading, and as competitors catch up (for example, the widespread praise of Google’s Gemini 3), it’s getting harder for OpenAI to lean on the old “cute feature you’ve built, but our model is still the best” narrative.
So what does all of this actually tell us?
For a moment it felt like ChatGPT was unbeatable, as if the AI match was already over. But now — after we’ve collectively picked our jaws off the floor from that first AI shock — it seems equally obvious that what we saw back then was only round one.
And be honest: how many times have you been asked by an older relative, “So… how do you use this AI thing?” And how easy was that to explain to someone completely outside the IT bubble?
Yes, you’ve been asked. And yes, the answer was fairly complicated: “Download this app, create an account, you only get X number of requests per day…” Even if said relative manages the process, they still have to be extremely motivated to turn it into a daily habit.
Microsoft made a brilliant move by betting on OpenAI early, but so far they’ve focused on embedding the models into their existing workplace tools. Makes sense — that’s where their distribution lives. Meanwhile, Google has been quietly but efficiently working in the background. They never needed to invent a new interface for their model. Instead, they’re slowly weaving Gemini into everything you already use in your daily life — your phone, search, email, maps, car, home.
And suddenly you don’t have to choose an AI service anymore.
It’s just… there.
And maybe that’s where the next round of the AI race will be decided — in distribution, not in the model itself.
