Top News

At Google I/O 2026, Sundar Pichai unveiled what may be Google’s most ambitious AI roadmap yet, signaling a shift from AI-powered features to fully agentic computing experiences. The keynote focused on how artificial intelligence is now deeply embedded across Google’s entire ecosystem, from Search and Workspace to Android, YouTube, Cloud, and developer platforms.

The company described the last year as a period of “hyper progress,” driven by rapid advances in AI models, infrastructure, and products used by billions of people every day. Ten years after declaring itself an “AI-first” company, Google says it is now entering a new phase where AI is no longer experimental, but central to how people search, create, work, and interact with technology.

AI Adoption Is Exploding

One of the clearest themes from Google I/O 2026 was scale.

Google revealed that it is now processing more than 3.2 quadrillion tokens per month across its AI-powered products and services, a dramatic jump from 480 trillion tokens at I/O 2025 and just 9.7 trillion two years ago. The company also shared that:

• More than 8.5 million developers are building applications using Gemini models every month

• Google’s APIs process roughly 19 billion tokens every minute

• Over 375 Google Cloud customers processed more than one trillion tokens each over the last year

These numbers reflect the accelerating adoption of AI tools by consumers, developers, and enterprises alike.

Gemini Powers Google’s Biggest Products

Google highlighted how Gemini AI models are now driving engagement across nearly all of its major products.

Search Gets More Conversational

Google Search continues to be at the center of its AI strategy. According to the company:

• AI Overviews now has more than 2.5 billion monthly active users

• AI Mode, Google’s conversational search experience, has surpassed 1 billion monthly active users

Google says Search is evolving from a traditional query-based engine into something that feels more like an ongoing conversation. Users are increasingly asking longer, more complex questions and interacting with Search in a more natural way.

The company also announced new “information agents” for Search. These agents will work continuously in the background, helping users track topics, monitor information, and take actions automatically.

Google is also introducing dynamic AI-generated interfaces directly inside Search. Using Gemini 3.5 and Google’s Antigravity platform, Search will soon generate custom layouts, visuals, and even persistent mini dashboards tailored to individual queries.

Gemini App Doubles in Size

The Gemini app has seen massive growth over the past year.

At I/O 2025, the app had 400 million monthly active users. In 2026, Google says the app has crossed 900 million users, while daily AI requests have increased more than sevenfold.

Google also revealed that over 50 billion images have been generated using its Nano Banana image generation models, showcasing the growing role of AI creativity tools.

Ask YouTube Reinvents Video Search

One of the standout announcements was Ask YouTube, a new AI-powered search experience for YouTube.

Instead of forcing users to manually browse videos, Ask YouTube can answer complex questions conversationally and jump directly to the most relevant parts of videos. The feature aims to make learning from YouTube faster and more intuitive.

For example, users can ask detailed questions like how to teach a child to ride a bicycle, and the system surfaces precise video segments instead of full-length videos.

Google says Ask YouTube will roll out broadly in the U.S. this summer.

Docs Live Brings Voice-Powered Productivity

Google also introduced Docs Live, a new Gemini-powered Workspace feature that allows users to create documents using natural speech.

Rather than typing structured prompts, users can simply speak naturally while Gemini organizes thoughts into a polished document. Google says future versions will support complete voice-based editing and document management.

Voice capabilities are also expanding to Gmail and Keep later this year.

Massive Infrastructure Investments

To support its expanding AI ecosystem, Google announced a dramatic increase in infrastructure spending.

The company expects annual capital expenditures to reach approximately $180 to $190 billion this year, nearly six times what it spent in 2022.

A major part of that investment centers around Google’s custom Tensor Processing Units (TPUs).

TPU 8t and TPU 8i

Google introduced its eighth-generation TPU architecture featuring two specialized chips:

• TPU 8t for large-scale AI training

• TPU 8i optimized for inference workloads

Google says TPU 8t delivers nearly three times the compute power of its previous generation and can scale training workloads across more than one million TPUs globally using JAX and Pathways.

Meanwhile, TPU 8i focuses on low latency and energy efficiency, delivering up to twice the performance-per-watt compared to earlier generations.

Gemini Omni: AI That Understands Every Modality

Google also introduced Gemini Omni, a new multimodal model capable of generating outputs across multiple formats from almost any input type.

The first release, Gemini Omni Flash, initially focuses on video generation but will later expand to image and text outputs. The model combines Gemini’s reasoning capabilities with Google’s generative media systems like Veo and Lyria.

Google says Omni Flash will be available across:

• Gemini app

• Google Flow

• YouTube Shorts

• APIs for developers and enterprises

Gemini 3.5 Flash Prioritizes Speed and Action

A major technical announcement at I/O 2026 was Gemini 3.5 Flash.

Google positioned the model as a breakthrough in balancing high intelligence with extremely fast output speeds. According to the company, Gemini 3.5 Flash is:

• Faster than competing frontier models

• Four times quicker in token output speed

• Strongly optimized for coding, reasoning, and real-world tasks

Google says the model is already transforming internal development workflows through its “agent-first” development environment called Antigravity.

The company also emphasized cost efficiency, claiming enterprises could save billions annually by shifting workloads to Gemini 3.5 Flash compared to more expensive frontier AI models.

Antigravity 2.0 and the Rise of AI Agents

Perhaps the biggest strategic shift announced at I/O was Google’s move toward AI agents.

Google introduced Antigravity 2.0, a new platform for building and orchestrating autonomous AI agents capable of performing long-running tasks independently.

This powers the new Gemini Spark, described as a personal AI agent that can operate 24/7 on behalf of users.

Gemini Spark can:

• Perform background tasks

• Integrate with Google services

• Work through Chrome and Android

• Eventually connect with third-party tools through MCP support

Google says Spark runs on dedicated cloud-based virtual machines and can continue operating even when users are offline.

AI Transparency and SynthID Expansion

With concerns around deepfakes and AI-generated media growing, Google expanded its SynthID watermarking technology.

The company says SynthID has now watermarked:

• Over 100 billion images and videos

• More than 60,000 years of audio content

Google is also expanding Content Credentials verification across Search and Chrome, helping users identify whether content originated from AI tools or traditional cameras.

In a notable industry collaboration, Google announced that companies including OpenAI, NVIDIA, Kakao, and ElevenLabs are adopting SynthID standards.

AI Comes to Wearables and Science

Google also previewed new intelligent eyewear powered by Gemini. The glasses will provide spoken assistance, navigation, messaging, and contextual information without requiring users to pull out a smartphone.

Audio-based smart glasses are expected to launch later this fall.

In addition, Google introduced Gemini for Science, a suite of AI tools designed to accelerate scientific research by integrating Gemini with major scientific databases and research systems.

The Beginning of Google’s Agentic Future

Google I/O 2026 made one thing clear: the company believes the future of computing will revolve around autonomous AI agents that can reason, act, and assist continuously across every surface.

From Search and Workspace to Chrome, Android, YouTube, and Cloud, Google is embedding Gemini deeply into its ecosystem while building the infrastructure required to support AI at unprecedented scale.

Rather than presenting AI as a standalone feature, Google’s vision at I/O 2026 was about transforming the internet itself into a more conversational, proactive, and agent-driven experience.