Google unveiled a sweeping expansion of its artificial intelligence strategy at its I/O 2026 developer conference, positioning the company for what Chief Executive Officer Sundar Pichai described as the next phase of computing: autonomous AI agents capable of handling increasingly complex tasks on behalf of users.
The announcements underscore the mounting competition among technology companies racing to commercialize generative AI tools while building the infrastructure needed to support them. Google said it expects to spend roughly $190 billion this year, largely on data centers and custom silicon, as it scales the systems powering its Gemini AI models.
“Gemini 3.5 and Antigravity are unlocking a new world of agents and agentic capabilities,” Pichai said during the company’s keynote presentation in Mountain View, California.
The company framed the event as a milestone in its decade-long transition to becoming an “AI-first” organization, a strategy Google says now spans chips, cloud infrastructure, research labs, developer tools and consumer products.
AI Adoption Accelerates
Google highlighted the rapid growth of AI usage across its ecosystem. The company said its systems now process more than 3.2 quadrillion tokens per month, up from 480 trillion a year ago and 9.7 trillion two years ago.
More than 8.5 million developers are building with Google’s AI models each month, according to the company, while Google APIs process roughly 19 billion tokens every minute.
The growth is also reshaping Google’s consumer products.
AI Overviews in Search now reaches 2.5 billion monthly active users, while the company said AI Mode has surpassed 1 billion monthly users within a year of launch. Google said the Gemini app now has more than 900 million monthly active users, more than double the figure from a year earlier.
Executives emphasized that users are increasingly engaging with conversational interfaces rather than traditional search formats, signaling a broader shift in how consumers interact with information online.
New Consumer AI Features
Google introduced several AI-powered tools designed to bring conversational interactions deeper into its products.
Among the new offerings:
- Ask YouTube, which allows users to search videos conversationally and jump directly to relevant segments.
- Docs Live, a voice-driven document creation tool enabling users to dictate and edit documents conversationally.
- Expanded AI capabilities inside Gmail, Keep and Maps.
The company also previewed a range of experimental products, including AI-powered smart glasses, collaborative creative tools and “Information Agents” in Search designed to continuously gather information and complete tasks in the background.
Gemini 3.5 Flash
At the center of the announcements was Gemini 3.5 Flash, Google’s latest AI model aimed at balancing performance, speed and cost efficiency.
Google said the model outperforms Gemini 3.1 Pro across several benchmarks while operating four times faster than competing frontier systems. The company also said the model can deliver advanced AI capabilities at less than half the price of comparable offerings.
The release reflects growing pressure across the industry to reduce the cost of deploying AI systems while maintaining competitive performance.
Gemini 3.5 Flash is available immediately through Google products and APIs, while Gemini 3.5 Pro is expected to launch next month.
Antigravity and AI Agents
Google also expanded its “Antigravity” platform, a system designed to coordinate and manage groups of autonomous AI agents.
The updated Antigravity 2.0 desktop application serves as a central interface for interacting with AI agents capable of carrying out long-running workflows.
One of the flagship agent products, called Gemini Spark, is designed to run continuously in the cloud, handling tasks across email, chat and web browsing. Google said the service will begin beta testing next week for subscribers to its AI Ultra plan.
The company also announced “Agentic Search,” a feature that enables AI agents inside Search to continuously monitor topics, collect information and perform actions on behalf of users.
Infrastructure Arms Race
Supporting the expansion is a massive increase in computing infrastructure.
Google unveiled two new custom AI chips, TPU 8t and TPU 8i, designed separately for training and inference workloads. The company said TPU 8t delivers triple the raw computing power of its predecessor and can scale across more than 1 million TPUs globally.
The investment reflects the escalating capital demands of the AI race, as companies including Microsoft, Amazon and Meta Platforms spend aggressively on data centers and specialized processors.
Google said both chips offer up to twice the performance-per-watt of earlier generations, a key metric as energy consumption becomes a growing concern across the industry.
Expanding Beyond Search
The announcements signal Google’s broader ambition to evolve beyond traditional search and become a central operating layer for AI-powered digital experiences.
In addition to consumer tools, the company previewed “Gemini for Science,” an experimental platform connecting AI systems to more than 30 life sciences databases to accelerate scientific research.
Other experimental projects included Google Flow, a collaborative workspace assistant, and Google Pics, an AI-powered image editing platform.
The breadth of the announcements highlighted Google’s effort to integrate AI across nearly every layer of its business as competition intensifies in the generative AI market.
Alphabet shares have faced increasing scrutiny from investors over whether AI-driven conversational interfaces could disrupt Google’s core advertising business. The company’s latest strategy suggests it intends to position AI not as a threat to Search, but as the next evolution of it.