Google’s I/O 2026 developer conference delivered a sweeping set of AI announcements, headlined by the introduction of Gemini Spark, a personal AI agent that runs continuously within Google apps, and Gemini Omni, a multimodal model designed to generate video from nearly any input. The event underscored Google’s aggressive push to embed agentic AI across its ecosystem and compete in the race toward multimodal intelligence.
DeepMind CEO Demis Hassabis described Gemini Omni as “our new model that can create anything from any input,” combining Gemini’s reasoning with media-generation systems including Veo, Nano Banana, and Genie. The first iteration, Gemini Omni Flash, will launch through Flow (Google’s AI filmmaking platform) and Flow Music. Demos showed the model generating a claymation-style educational video and enabling conversational editing that keeps characters and backgrounds consistent even after changes—addressing a common challenge in AI video generation. Hassabis framed Omni as a “step towards artificial general intelligence,” noting Google has spent a year building a world model AI that can simulate the world.
Simultaneously, Gemini Spark marks a leap into daily automation. The 24/7 agent lives inside Gmail, Docs, and Meet, proactively scanning emails for hidden fee changes, creating documents from meeting notes, flagging unusual credit card charges, and executing other recurring tasks. It will be available to testers immediately, with broader beta access next week for U.S. subscribers of Google’s new AI Ultra tier ($100/month). Google also unveiled Gemini 3.5 Flash, its fastest model to date, claiming output tokens are generated four times faster than other frontier models—ideal for coding and agent-based workflows.
Search received its biggest upgrade in 25 years. An expanded search box accepts longer, more detailed queries, while new AI-powered search agents can comb blogs, news sites, and social media, then send periodic updates to users. AI Mode, introduced last year, now has over 1 billion monthly users, with query volumes more than doubling each quarter. The Gemini app itself hit 900 million users, up from 400 million at the previous I/O.
Beyond software, Alphabet’s cloud division posted $20 billion in Q1 revenue—a 63% year-over-year jump—and carries a $460 billion backlog. A new AI infrastructure venture with Blackstone aims for 500 megawatts of capacity by 2027. Google will sell its eighth-generation custom chips directly to some customers. AI smartglasses designed with Warby Parker and Gentle Monster, capable of directions, translations, and DoorDash ordering, are slated to ship this fall. Alphabet stock (GOOGL) has risen roughly 25% year-to-date, pushing its market valuation near $5 trillion.
