.jpg)
Google I/O 2025 was exactly what it needed to be: a performance, not a surprise. There was no reinvention, no late-breaking moonshot. Instead, it was a well-choreographed declaration that Gemini is no longer a lab experiment or a PR tool. It’s the operating system now.
It’s not every day that a keynote makes you feel like you’re watching a company calmly bolt AI into the frame of everyday life, but that’s what this year’s I/O managed. Gemini is now deeply stitched into Chrome, Search, Android, and Workspace. This isn’t the clunky bolted-on assistant era of yore. This is Google saying, very clearly, that everything it makes will be made better by generative AI. Or at least, that’s the plan.
The mood wasn’t jubilant, but it was confident. AI Overviews have started rolling out as the default in Search for US users, which, if extended to India, would mark a fundamental shift in how performance marketing and organic discovery even work. It would also render half the SEO playbooks obsolete overnight. The Project Astra demo, meanwhile, felt like the kind of demo you’d expect in a concept reel for a Black Mirror episode that hasn’t yet been made—real-time visual understanding, voice interactivity, object permanence, and context awareness, all in one seamless clip. There’s still a question of how much of it works outside a lab environment, but the direction of travel is clear.
Then there was Jules, the AI coding assistant, which seems to be Google’s answer to the developer allegiance OpenAI has been steadily cultivating. If you’re a coder and haven’t picked a side, Google just gave you one. Add to this Flow and Lyria—Google’s video and music generation tools—and what we saw was not just an update to the Gemini stack, but the beginnings of a fully generative creative suite.
And then came the glasses.
Google’s new smart glasses, developed in partnership with Warby Parker, Gentle Monster, and Samsung, are less a product announcement and more a signal: the company is serious about XR. There’s no formal launch, no pricing, not even a retail roadmap yet. But the hardware exists, and it works. Or at least, it works well enough to be shown off.
Here’s the thing, though. The idea of smart glasses is still haunted by the ghosts of past optimism. Google Glass. Snap Spectacles. Ray-Ban Meta. The graveyard is well-populated, and the use case still feels more utopian than inevitable. The glasses Google showed off this year were better—lighter, smarter, more grounded—but they’re still glasses. And people still don’t really want to wear hardware on their face to talk to an assistant that lives in their phone.
Also worth noting: the XR reveal carried a strong scent of strategic counterprogramming. Meta has staked its entire consumer AI narrative on wearable tech. Google, by showing off its own glasses, gets to participate in that conversation without fully committing to its premise. It’s a move that says: we can do this too, but we’re not betting the company on it.
Which brings us to the part of the program that Google didn’t schedule, but which has become something of an annual tradition. The Sam Altman disruption.
Just as Google was rolling out I/O to hundreds of journalists and live streamers, OpenAI casually dropped a new model—GPT-4.1—into ChatGPT, upgraded the Codex coding agent into a full-blown terminal CLI, and, most dramatically, unveiled a Bloomberg video tour of a $500 billion AI infrastructure initiative named Stargate.
There were no live demos. No fanfare. Just Emily Chang, a hard-hatted Sam Altman, and the quiet suggestion that OpenAI wasn’t just building tools—it was building the substrate on which the future might run.
The contrast couldn’t be clearer. Google is shipping interfaces. OpenAI is building the walls of a new temple. One company is betting on its ecosystem. The other is betting on its mythology.
That’s not a knock on Google. In fact, this I/O was a reminder that scale and integration are still the company’s strongest weapons. Everything Google announced was actually usable. Brands can build with this stuff. Developers can deploy it. Users will interact with it, likely without knowing that Gemini is even there. That’s a win.
Google’s I/O wasn’t just a product launch—it was a study in how to maintain tempo in a race that’s become metaphysical. The company did everything right. It showed capability, control, vision. It held the centre. But OpenAI didn’t need a stage. It just needed timing.
.jpg)
