Google quietly made the phone optional—and I think that’s the biggest story of I/O 2026

There’s a moment during the Google I/O 2026 keynote that I keep coming back to. Josh Woodward is on stage demoing Gemini Spark, and he says, almost as an aside, “You can close your laptop.” Then, a little later, he pulls out his phone, fires off a voice command stacking three separate tasks, and puts it face-down on a table. He doesn’t pick it up again. The crowd barely reacts. But that moment—a phone sitting idle on a stage, doing nothing, while Gemini does everything—is the most consequential thing Google showed.

I don’t think they meant to announce the death of the upgrade cycle. But I think they did it anyway.

The brain no longer lives in your device

Here’s the thread I can’t stop pulling. Gemini Spark runs on dedicated virtual machines on Google Cloud. Not on your phone. Not on your laptop. On Google’s infrastructure, 24/7, whether your devices are on or off or at the bottom of your bag. The macOS app—built in under 100 days by a small team using Antigravity—brings the same intelligence to a desktop.

Android Halo is coming as a dedicated home base for agents on your phone, but notice the framing: your phone becomes the home base for agents, not the source of their intelligence. The glasses, demoed with Nishtha on stage, ordered a nitro cold brew from DoorDash, added a dinner to a calendar, and summarized muted texts. Her phone was in her pocket, screen off, essentially a relay station for a transaction that Gemini negotiated entirely on its own.

What all of this describes is a brain that has moved out of your pocket.

The end of the silicon arms race

For 15 years, the smartphone upgrade cycle ran on the promise that more silicon means a smarter experience. A better camera sensor, a faster neural processing unit, a bigger battery to power more on-device inference. The phone was the platform. The phone was the computer. Everything meaningful happened inside it, which meant you had an incentive to upgrade it every two or three years, because a faster phone was a meaningfully better experience.

That logic is breaking down, and Google I/O 2026 is the clearest signal yet of how fast it’s happening.

When Spark is handling your RSVP list, drafting follow-up emails to neighbors, compiling your kids’ end-of-year checklist, and prepping a block-party deck in Google Slides—all from a single voice command, running in the background on cloud VMs—the question “what phone are you running?” becomes almost quaint.

The A-series chip in your two-year-old Pixel isn’t the bottleneck. Gemini 3.5 Flash, running somewhere in a Google data center at nearly 1,500 tokens per second, is doing the cognitive work. Your handset is the microphone, the screen, and the radio. It’s thin-client computing, and we’ve been so dazzled by what the AI can do that we haven’t stopped to notice the hardware has been demoted.

Your phone just became an interface

The glasses make this unavoidable to think about. Nishtha’s Gentle Monster frames ordered her coffee before she touched anything. The intelligence wasn’t in the glasses—those are audio hardware with a camera and a Bluetooth connection. I found the intelligence in Gemini, in the cloud, peering through the camera, consulting her Calendar, opening DoorDash, clicking through option screens on the phone in her pocket that she never touched. The glasses were a window. The phone was a proxy. Neither was the brain.

From computers to sensor arrays

So what are you actually buying when you buy a phone today?

You’re buying a sensor array. A camera that feeds Gemini’s vision. A microphone that feeds its ears. A screen for the moments when you want to see something rather than hear it. A battery that keeps the radio alive. These aren’t nothing—camera quality still matters enormously, and battery life is load-bearing for everything else. But they are different from what we used to buy. We used to buy processing power. Now we’re buying connectivity quality and form factor.

The flagship problem nobody wants to discuss

This has uncomfortable implications for the premium end of the market. The $1,200 flagship pitch has always been justified by the performance gap between tiers. That gap doesn’t disappear, but it narrows in weird ways. If most of your cognitive heavy lifting is outsourced to the cloud, you don’t need the fastest NPU. You need a good enough connection and a camera lens that doesn’t embarrass you. Mid-range hardware running Gemini Spark is going to feel remarkably close to flagship hardware running Gemini Spark, because Spark doesn’t care what’s in your hand.

Apple and Google are betting on different futures

I’m not saying phones stop mattering. I’m saying the reason they matter is shifting, and the industry hasn’t caught up to that conversation yet. The marketing still sells you on the chip. The experience is increasingly selling you on the network. Apple, notably, has been resistant to this model—their intelligence is deliberately more on-device, which is a coherent strategy but also a different bet about where the world is going. Google has placed its chips, loudly, on the cloud.

The upgrade question has changed

And if Google is right, then the question for the next upgrade cycle isn’t “Is this phone faster than my last one?” It’s “Does this phone stay out of the way well enough for Gemini to do its job?” That’s a lower bar. And a lower bar means longer replacement cycles, which means the economics of the handset business get quietly worse for everyone selling handsets—including Google itself.

Did Google just disrupt its own hardware business?

They may have just disrupted their own hardware division while the audience applauded a party planning demo. That’s either genius or an accident. I cannot tell which, and I’m not sure it matters.

What I do know is that I looked at my phone differently after watching that keynote. Not with less affection. Just with less reverence. It’s a beautiful piece of glass that asks Gemini for help on my behalf. The interesting part of computing has moved somewhere else.