Google I/O just killed your startup
Let's take a moment to step inside the world of a startup founder in 2025.
You raise $100 million, hire a team, obsess over the product, and launch something great.
Then a giant tech company drops a new feature that does the same thing: except it’s integrated into a suite of products already used by billions, maybe even offered for free.
This week, that's exactly what just happened to entire categories of startups, all at once.
Don't believe me? Let's take a look at what was announced at the Google I/O event:
Real-time translation on video calls
Google Meet now does real-time voice translation, and it keeps your voice, tone, and timing intact.
Initially it will support English and Spanish, with more languages coming soon.
RIP to... KUDO
KUDO raised $28 million to bring live interpretation to enterprise video calls.
They built a strong product, but Google just turned the feature into a default setting.
AI-generated video with native audio
Google has introduced a new version of Veo, its video generation model. It now creates 1080p clips with ambient sound, dialogue, and audio effects baked in.
This is the first model of its kind that combines high-quality video and audio in a single output.
RIP to... Runway
Runway has raised over $300 million and has long led the charge on AI-generated video.
But its models still don’t handle audio. Google just leapfrogged them.
(I'm exaggerating to say that Runway is dead - their video model is still excellent and we used it a huge amount in our work at Topham Guerin. But the pressure is on for them to overtake Veo 3 before customers jump ship.)
AI-powered video editing
Flow is Google’s new creative tool built into YouTube Studio. It uses Gemini to help creators script, storyboard, and edit videos faster.
The goal is clear: turn your idea into a Short in minutes, without leaving the platform.
RIP to... Captions
Captions raised $60 million to bring AI editing tools to creators.
Now Google has built the same tools directly into the biggest video platform on Earth.
Smarter, safer password management
Chrome and Android are becoming full-service sign-in assistants tjat:
- Generate strong passwords
- Flag compromised ones
- Fix issues automatically
- Seamlessly log you in to web and apps
RIP to... 1Password and Dashlane
Between them, they raised over $800 million to manage passwords.
But now Google is giving that away for free, built into tools people already use every day with a product that is easily a generation ahead.
I use NordPass and after this demo am already considering cancelling my subscription.
Virtual try-ons with your own photo
Upload a photo and Google will show you how clothes will look on your body. The system accounts for shape, pose, fabric, and texture.
It's live now in Search Labs for US users.
RIP to... Every ecom startup pitching virtual try-ons, including Fashable who seem to have already deactivated their website
Fashable raised $6M to build virtual try-on tools, but Google just shipped a version that works at global scale, across billions of listings, directly inside Search.
It's a tough time to be building in this space.
So what now?
We have looked at five examples, and you could easily list a dozen more just from the Google I/O event.
Real-time translation. AI video generation with audio. AI video production studio. Agentic login. Smarter shopping.
All companies worth investing in, now just another Google feature.
And you could repeat the exercise with recent events from Microsoft and OpenAI and knock out another dozen startup categories.
Now, I exaggerate when I say these product categories are wiped out.
If some of these startups move fast to leapfrog Google or go deep on specific niches, they might hold their own.
Competing on price, UX, or meeting the needs of a unique customer segment are viable options.
The deeper problem
This isn’t just a platform advantage. It’s a fundamental structural challenge of building a product offering built around AI.
Owning a distribution platform now matters more than building a perfect product. When you’ve got the right model and data stack, launching something new is the easy part.
And that’s the real issue for venture-backed startups. VC relies on winner-take-all outcomes. You need a few portfolio companies to dominate their category.
But what happens when the “winner” is just a feature inside someone else’s ecosystem?
As far as VCs are concerned, it's not enough for their portfolio companies to survive or even thrive - they need to dominate the market entirely to justify the investment.
That is just going to be more difficult than ever in an age of AI -powered product development
Because if distribution is the moat, and the platforms keep pulling features closer to the metal, then the number of breakout challengers might be smaller than anyone wants to admit.