In May 2024, Google's AI told people to put glue on their pizza and eat rocks. The company was a punchline. ChatGPT had stolen its thunder, and Silicon Valley consensus was that the search giant had fumbled the most important technology shift in a generation.

Two years later, that consensus has flipped entirely.

At Google I/O 2026 on Tuesday, CEO Sundar Pichai revealed that Gemini now has 900 million monthly active users, more than doubling in a single year. That number matches OpenAI's self-reported active user count for ChatGPT and dwarfs Anthropic's Claude by roughly 30x in estimated web traffic. The New York Times published a piece the same day headlined: "How Google Is Starting to Win the A.I. Race."

But the user number was just the opening act. Google dropped three major product announcements that collectively represent the most aggressive AI push in the company's history.

Gemini 3.5 Flash: Frontier Intelligence at a Third of the Price

Gemini 3.5 Flash is Google's new default model across the Gemini app, Search, and the Gemini API. According to Google, it surpasses Gemini 3.1 Pro in coding, agentic tasks, and multimodal benchmarks while running 4x faster than comparable frontier models in output tokens per second. It costs between a third and a half less than competing frontier models.

Pichai told reporters: "You no longer have to trade quality for latency." The model is rolling out starting today. Gemini 3.5 Pro, the heavier version, is being used internally and will ship next month.

Gemini Spark: A 24/7 Personal AI Agent

Google also announced Gemini Spark, a general-purpose AI agent that lives inside the Gemini app and can reason across connected applications. Unlike existing chatbot features, Spark is designed to take actions on your behalf: booking, researching, managing tasks, and navigating your digital life autonomously.

Spark enters beta next week for trusted testers and Google AI Ultra subscribers. Mashable called it Google's answer to OpenAI's operator products and emerging personal AI agent platforms. The difference: Google has distribution that nobody else can match. Gemini is already on every Android device. When Apple integrates it as the foundation for a future version of Siri (a deal announced in January), Gemini will effectively be baked into virtually every smartphone on the planet.

Gemini Omni: Creating Video From Any Input

The third announcement was Gemini Omni, a new model that combines reasoning with creation. It can generate video output from images, audio, video, or text input. Google is positioning it as a universal creative engine, not just a language model.

The Money Story

Here is where Google's advantage becomes structural. Unlike OpenAI and Anthropic, which are still losing billions operating expensive data centers, Google is already making money from AI. Its advertising revenue rose 16% last quarter to $77 billion, fueled by AI tools that help marketers collect deeper data on user interests. The company also introduced a new AI-powered shopping cart feature at I/O designed to increase conversion rates for online retailers.

OpenAI and Anthropic are both preparing for IPOs later this year, with combined losses still running north of $30 billion annually. Google is posting record profits while shipping comparable (and in some benchmarks, superior) models.

The Redesign Nobody Expected

Google also rolled out "Neural Expressive," a full redesign of the Gemini app across Android, iOS, and web. The new interface includes fluid animations, a pill-shaped prompt box, inline Gemini Live voice conversations (no more switching to a fullscreen interface), and responses designed to surface the most important information visually with inline images, timelines, and interactive elements. The redesign also introduced a two-tier subscription model: a new $100/month AI Ultra plan with 5x usage limits sits alongside the existing $20/month Pro plan.

What This Means

Two years ago, the narrative was that Google was too big, too slow, and too cautious to compete. That narrative is dead. The company has the users (900 million), the revenue ($77 billion in advertising alone last quarter), the distribution (every Android phone, soon every iPhone), and now the models to compete on benchmarks.

The question is no longer whether Google can keep up. It is whether anyone can keep up with Google once it starts shipping to 4 billion devices simultaneously.

Google I/O 2026 continues through Wednesday with developer sessions on API details, enterprise integrations, and shipping timelines.

Sources: CNBC (Jennifer Elias, May 19), The New York Times (Brian X. Chen, May 19), 9to5Google (Abner Li, May 19), Engadget (Kris Holt, May 19), Mashable (Timothy Beck Werth, May 19).