Google's AI momentum is undeniable, but the company faces a critical gap in one of the hottest areas of artificial intelligence. At the company's flagship I/O developer conference on May 20, CEO Sundar Pichai showcased impressive new AI products and staggering adoption numbers, yet also revealed that Google is still playing catch-up in AI-powered coding compared to rivals like Anthropic and OpenAI.

What Did Google Announce at I/O 2026?

Google rolled out a slate of new AI tools designed to bring its vision of a super-intelligent assistant closer to reality. The company introduced Spark, a new AI agent that can connect to email and execute tasks without requiring users to keep their laptops open. Google also unveiled Gemini Flash 3.5, a new AI model optimized specifically for AI agents and designed to be cost-effective.

Perhaps most significantly, Google announced a major overhaul of its core product: Google Search. The company is now integrating its latest AI features directly into the search box while allowing new AI agents to search for users in the background. These moves underscore Google's strategy to embed AI deeper into the products billions of people use every day.

Why Are Google's Adoption Numbers So Staggering?

The numbers Pichai shared painted a picture of explosive growth in AI adoption across Google's ecosystem. According to the CEO, people worldwide now consume 3.2 quadrillion AI tokens per month using Google's products, a sevenfold increase from just one year earlier. To put that in perspective, an AI token is roughly equivalent to a word or small piece of text; this metric shows how intensively users are engaging with Google's AI tools.

The Gemini app, Google's standalone AI chatbot, recently surpassed 900 million monthly active users, more than double the 400 million users it had approximately one year ago. Additionally, AI Overviews in Google Search, which provide AI-generated summaries alongside traditional search results, now reach over 2.5 billion monthly active users.

• Gemini App Growth: Reached 900 million monthly active users, up from 400 million roughly one year prior
• AI Token Consumption: Users now process 3.2 quadrillion tokens monthly, a sevenfold increase year-over-year
• Search Integration: AI Overviews in Google Search serve over 2.5 billion monthly active users with AI-generated summaries

What's the Catch? Google's Coding Problem

Despite the impressive adoption metrics, Google faces a notable weakness in a field that has become central to AI competition: coding. Multiple developers interviewed at the conference confirmed that Gemini's coding abilities lag behind competitors like Anthropic's Claude Code and OpenAI's Codex. This gap is particularly significant because AI-powered coding tools have become a major battleground among AI companies, with enterprises and developers increasingly relying on these assistants to write and debug software.

Google is acutely aware of the problem and is working hard to close it. The company heavily showcased Antigravity, its AI agent development product, at I/O as part of its effort to improve coding capabilities. According to Varun Mohan, a DeepMind engineer who cofounded the AI coding startup Windsurf, Google's models "are continuing to improve," which is supercharging its coding abilities as well.

"Google's models are continuing to improve, which is supercharging its coding abilities as well," said Varun Mohan, DeepMind engineer and cofounder of the AI coding startup Windsurf.

Varun Mohan, DeepMind Engineer and Cofounder, Windsurf

Why Did Pichai Ask for More Time on Gemini 3.5 Pro?

One of the most anticipated announcements at I/O never came. Many in the industry expected Google to launch Gemini 3.5 Pro, its latest frontier AI model, at the conference. Instead, the company's newest flagship model remains unavailable, drawing audible groans from the audience. Pichai acknowledged the disappointment and asked for patience, saying "I know you can't wait to get your hands on it." He requested that the industry give Google until next month to deliver the model.

Pichai

This delay is noteworthy because the AI industry has been closely watching Google's big models, especially after Gemini 3 exceeded expectations and solidified the company's position as a serious AI competitor. The postponement suggests that Google may be taking extra time to ensure the model meets its quality standards or to address specific performance gaps, particularly in areas like coding where competitors have pulled ahead.

How to Stay Updated on Google's AI Roadmap

• Monitor Official Announcements: Follow Google's official blog and developer documentation for updates on Gemini 3.5 Pro and other AI model releases
• Track Adoption Metrics: Watch for quarterly reports on Gemini app users and AI token consumption to gauge how quickly Google's AI products are being adopted
• Follow Competitive Benchmarks: Compare Google's coding model performance against Claude Code and other competitors through published benchmarks and developer reviews

What Does This Mean for the Broader AI Race?

Google's I/O presentation revealed a company that is simultaneously winning and struggling. The company's massive spending on AI is clearly paying dividends in terms of user adoption and product integration. However, the delay of Gemini 3.5 Pro and the acknowledged lag in coding capabilities suggest that Google's competitors are not standing still. The company's AI chief, Demis Hassabis, declared that humanity now stands at the "foothills of the singularity," the theoretical point at which AI begins to improve itself and surpasses human intelligence. Yet even as Google positions itself as a leader in this transformation, it must first catch up in specific technical areas where rivals have gained ground.

For developers, enterprises, and users, the takeaway is clear: Google's AI ecosystem is growing rapidly and becoming more integrated into everyday tools. However, those specifically looking for cutting-edge coding assistance may still find better options elsewhere, at least for the next month.