Sebastian Gomez
Chrome and the Web Reinvent Themselves at Google I/O 2025
Hello everyone. As many of you know, I had the privilege of being present at Google I/O, the epicenter of innovation where Google shows us the future. As a Google Developer Expert in Web Technologies, my radar was especially focused on everything new that Chrome and the web platform have to offer us. And there is plenty of news!
Status note (June 2026). This post is a recap of the announcements from Google I/O 2025, so its claims are time stamped by design. A year later, several APIs that were in Origin Trials or in Canary back then have already reached stable. I keep the original recap and add, where relevant, a "current status" note so you don't walk away with an outdated fact.
Beyond the spectacular AI announcements that grabbed the headlines (which I'll talk about another time), I want to share with you those gems that will directly impact our work as web developers, transforming the way we build online experiences.
1. AI becomes part of our tools: DevTools powered by Gemini
One of the announcements that excited me most was the deep integration of AI directly into Chrome DevTools. That's right! We now have Gemini assistance to optimize our debugging workflows. Imagine being able to apply AI suggested changes directly to the files in your workspace from the Elements panel. Or how about getting contextual insights into your Core Web Vitals performance in the revamped Performance panel, simply by asking the AI. This will not only save us time, it will democratize access to complex optimizations.
On top of that, Google keeps betting heavily on client side AI with the built in APIs that use Gemini Nano. Since Chrome 138 we have, in stable, the Summarizer, Language Detector, Translator APIs and the Prompt API for Chrome extensions. And at the time more were on the way, such as the Writer and Rewriter APIs (then in Origin Trials) and the Proofreader API with multimodal capabilities (then in Canary). This opens up a universe of possibilities for building smarter, more private, lower latency web apps.
Current status (June 2026). The Writer, Rewriter, and Proofreader APIs have moved on from their trial phase; before building a project on them, confirm the exact availability (web vs. extensions) in the official Chrome built in AI APIs documentation. The on device model is still Gemini Nano.
2. Modern UI with CSS: less JavaScript, more power
Those who know me know how much the evolution of CSS excites me. Google I/O brought us fantastic news on this front. If you want to dig deeper into these topics, I revisit them in my sibling posts from the CSS series on transitions and animations.
Carousels with pure CSS: At last! Creating interactive, attractive carousels is now possible with just a few lines of HTML and CSS, thanks to new CSS primitives that landed with Chrome 135. Pinterest, one of the first to adopt them, cut its code by 90 percent, going from 2000 lines of JavaScript to just 200 of CSS. Impressive! You can read the full case in the Chrome for Developers carousels with CSS article.
Interest Invoker API: this API, introduced as an origin trial, lets us show popovers declaratively when a visitor's interest is active for a short period. Combined with the Anchor Positioning API and the Popover API, we can create rich, interactive tooltips and hover cards without a single line of JavaScript.
Current status (June 2026). The Interest Invoker work (the interesttarget attribute) kept advancing in standardization since I/O 2025.
3. Instant navigations and cinematic transitions
User experience (UX) is key, and Chrome keeps pushing the limits to make it smoother.
Speculation Rules API: this API, which evolves on prefetch and prerender, lets developers tell the browser which pages are likely to be visited next. Chrome can then preload or even prerender these pages in the background, making navigation feel almost instant. It's especially powerful for multi page applications (MPAs) and offers a more expressive, configurable JSON syntax. We can define rules statically or inject them dynamically with JavaScript based on user interaction.
View Transitions API: although we already knew it for Single Page Applications (SPAs) since Chrome 111, the big news at I/O 2025 was its expansion to full page navigations, that is, cross document transitions for multi page applications (MPAs). This removes jarring jumps and gives our websites a native app feel, greatly improving the user's perception. You can define specific transitions for elements with the view-transition-name property in CSS. If you want to see practical examples, I cover it in my post on CSS transitions.
Current status (June 2026). Cross document View Transitions (MPA), which were the big news at I/O 2025, are already in stable in Chrome. They are no longer a future promise, you can use them today.
4. WebAssembly (Wasm) keeps maturing for heavy workloads
WebAssembly continues to be a pillar for bringing native performance to the web. Some highlighted advances were:
Memory64: lets Wasm linear memory exceed 4 GB, crucial for large AI models and demanding applications. This proposal has already reached stable.
Better interoperability with the web and WebGPU: makes it easier for C++ applications that run on device to run on the web with minimal modifications, taking advantage of the GPU.
Relaxed SIMD: performance improvements for vector operations, key in Machine Learning, with new instructions that can speed up workloads by 1.5 to 3 times, according to the data shared in the V8 Relaxed SIMD announcement.
5. Baseline: clarity on feature compatibility
Knowing which web features are safe to use across all browsers is a constant challenge. Baseline keeps growing and its compatibility information is now integrated into tools we use every day, like VS Code and linters like ESLint. On top of that, the web features dataset is fully mapped, which lets you see the Baseline status of each feature across the major browsers.
Looking to the future as web developers
From my perspective as a GDE, these updates are more than just new tools; they are enablers of the next generation of web experiences. AI integration in DevTools and in the browser itself will radically change our productivity and the capabilities of our applications. The improvements in CSS and the transition APIs will let us create richer, smoother interfaces with less effort. And the ongoing evolution of WebAssembly and Speculation Rules will keep driving performance.
Google I/O made it clear that the web is more alive than ever, and Chrome remains a key engine of this evolution. I encourage you to explore these new APIs, experiment with them and, as always, share your discoveries with the community.
3-point summary
- AI became part of both the DevTools (Gemini assistance) and the browser (APIs on top of Gemini Nano: Summarizer, Language Detector, Translator, and the Prompt API), and the Writer, Rewriter, and Proofreader APIs have kept advancing since.
- CSS gains ground: pure CSS carousels and the Interest Invoker API let us build rich UI with far less JavaScript.
- UX takes a leap with Speculation Rules for near instant navigations and cross document View Transitions, now in stable, while WebAssembly matures with Memory64 and Relaxed SIMD.
Suggested exercises
- Build a small Chrome extension that uses the Prompt API or the Summarizer API on top of Gemini Nano and summarize a text on device, without sending anything to a server.
- Build a carousel with only HTML and CSS using the new primitives, without writing any JavaScript, and compare how many lines you need versus a traditional library.
- Add View Transitions between two pages of a multi page site with the
view-transition-nameproperty and check the difference in how navigation feels.
That's all. I hope this Google I/O 2025 recap is useful to you and that it encourages you to try these APIs on a project you have in mind.
Which feature excites you most? Have you tried any of these tools yet? Leave me your comments below, and if you liked the post, remember you can also share it using the social links below. Good luck!
Sebastian Gomez
Creador de contenido principalmente acerca de tecnología.