Google’s annual developers event, the I/O 2026, kicked off on May 19 with major announcements around artificial intelligence (AI), Gemini upgrades, Search, and other platforms. However, the key highlight of the event was Google’s roadmap to enter the “Agentic era.”
During the keynote address, CEO Sundar Pichai emphasised the fast progress in AI, with major improvements in AI models, infrastructure, and products that are used by billions of people worldwide. He recalled the period of growth as “hyper progress,” which is impacting both technology and real-world applications at a large scale.
Now, Google is entering a new phase where AI is no longer just experimental or optional. But it has become a core part of everyday life. Here’s what the Google I/O 2026 announcements reveal about the company’s push toward a more agent-driven ecosystem.
Must read: Google redesigns the Gemini app and brings new agentic AI tools: Here’s what’s new
AI use grows at scale
Pichai revealed that the company is processing more than 3.2 quadrillion tokens per month, a massive jump from 480 trillion tokens per month at Google I/O 2025. Google claimed that the usage has grown exponentially as more products integrate AI features.
In addition, Google’s AI Overviews now has 2.5 billion monthly active users, and AI Mode has 1 billion monthly active users. For the Gemini app, its active users have grown from 400 million to 900 million users in just one year. Its Nano Banana tool also experienced massive growth, with people generating over 50 billion images worldwide.
With growing use and demand for AI, Google has committed to spending over $180 to $190 billion in infrastructure investment. However, the major part will be used for Google’s custom Tensor Processing Units (TPUs).
At I/O 2026, Google announced two eighth-generation TPU architectures: TPU 8t for large-scale AI training and TPU 8i for inference workloads.
Must read: Android 17 features revealed: What Google Pixel and Android users can expect
Gemini Omni: This is a new family of multimodal models specifically designed for AI video generation. Google claims that the model creates outputs in multiple formats from a wide range of input types. The model supports conversational editing, allowing users to edit characters, backgrounds, and other elements using voice commands.
Initially, Google is rolling out the first version of Gemini Omni Flash, which is designed for AI video generation, but Google plans to expand its capabilities later to support image and text generation as well. Read more here.
Gemini 3.5 Flash and Antigravity 2.0: Google launched its new generation flagship AI model, the Gemini 3.5 Flash, which is said to bring improved performance across coding, reasoning, and multi-model understanding. The company says that it offers 4 times faster performance and that it's faster than competing AI models. Read more here
Google also showcased a new version of its agentic coding app dubbed Antigravity 2.0, where users can run several specialised AI agents. With the app, users can create custom “subagent” workflows, automate tasks, and schedule processes to continue running in the background without constant user input.
Gemini app upgrades: Google also showcased a revamped Gemini app interface with smooth animations, vibrant colours, new typography and haptic feedback. The company calls it a “Neural Expressive design." The app gets upgraded performance with Gemini 3.5 Flash, AI video generation capabilities with Gemini Omni, Agentic features with Gemini Spark, and more.
Gemini Spark: Google also announced its new personal AI agent, Gemini Spark, that can run 24/7 on users' behalf. The agent can be integrated within Google Workspace tools to gain context, and it can run tasks in the background.
Google audio-based smart glasses: Google also previewed new audio-powered smart glasses at I/O 2026, which are said to debut later this year. To bring the smart glasses, the tech giant has partnered with Warby Parker and Gentle Monster, that works similar to Meta RayBan glasses. However, it will be powered by Google’s Gemini AI model.
Apart from these major announcements, Google also showcased Gemini for Science, bringing a set of AI tools for researchers. It also showcased new AI-powered features coming to YouTube, such as Ask YouTube for video search and Docs Live for seamless documentation generation. This year’s I/O was full of AI announcements, highlighting Google’s broader transformation toward an agentic AI ecosystem where intelligent systems can assist users, automate workflows, and work across apps, devices, and services with minimal human intervention.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。