What to expect from Google this week

The world’s most influential tech giant is about to step into the spotlight once again—but this time, it’s not just about flashy demos or incremental updates. As Google prepares to open the doors to its annual I/O developer conference in Mountain View, California, the company finds itself at a pivotal crossroads. Once the undisputed titan of artificial intelligence, Google now finds itself playing catch-up in a race it once led. While the world watches for the next big breakthrough in generative AI, the real story may lie not in who’s winning the coding wars—but in how Google is quietly reshaping the future of science, health, and human knowledge.

For years, Google has been synonymous with innovation. From search to Android, from YouTube to self-driving cars, the company has repeatedly redefined how we interact with technology. But in the age of foundation models—the massive AI systems powering everything from chatbots to drug discovery—Google’s position has shifted. A year ago, the narrative was different. With the launch of Gemini 2.5 Pro in March 2025, Google was riding high. Its models were competitive, if not dominant, and the distinctions between top-tier AI systems often felt academic. Today, the landscape has changed dramatically.

The Coding Crisis: Can Google Catch Up?

At the heart of Google’s current challenge is a quiet but urgent crisis: its AI coding tools have fallen behind. While Anthropic’s Claude Code and OpenAI’s Codex have surged ahead in performance, reliability, and developer adoption, Google’s offerings have struggled to keep pace. The gap has become so pronounced that internal reports suggest even Google’s own engineers at DeepMind—the company’s elite AI research division—have been granted special access to Claude Code to stay productive. It’s a humbling admission from a company that once prided itself on building the world’s best tools from the ground up.

This internal reliance on a competitor’s product is more than just an embarrassment—it’s a strategic vulnerability. If Google’s own researchers can’t rely on its AI coding assistants, how can it expect external developers to do so? The stakes are high. Coding is no longer just about writing software; it’s about accelerating innovation across industries. From fintech to biotech, AI-powered development tools are becoming the new operating system for progress. And right now, Google is missing out.

But the company isn’t sitting still. According to recent reports, Google has formed a new AI coding team within DeepMind, signaling a renewed commitment to the space. Even more telling is the involvement of John Jumper, the Nobel Prize-winning scientist behind AlphaFold, who is reportedly lending his expertise to the effort. Jumper’s work on protein folding revolutionized biology, and his shift toward coding tools suggests Google is betting on interdisciplinary breakthroughs. The expectation is that I/O 2025 will feature a major update to Antigravity, Google’s agentic coding platform—a system designed to not just autocomplete code, but to reason, plan, and execute complex software tasks autonomously.

Still, skepticism remains. Even with internal access to superior models, Google’s public releases have lagged. Unless there’s been a sudden, dramatic leap in capability, it’s unlikely that any announcement at I/O will immediately close the gap with Claude Code or Codex. The road back to the top will be long, and the competition isn’t standing still.

The Science Imperative: Where Google Still Leads

While the coding battle grabs headlines, Google’s most transformative work may be happening in a quieter, more profound domain: scientific discovery. For years, DeepMind has been pushing the boundaries of what AI can achieve in fields like biology, chemistry, and physics. The most famous example is AlphaFold, the AI system that solved the decades-old “protein folding problem”—a breakthrough that has accelerated drug discovery, vaccine development, and our understanding of disease.

But AlphaFold was just the beginning. Google’s AI for science initiatives now span climate modeling, materials science, and even quantum computing. At I/O, expect to see deeper dives into how AI is being used to simulate complex systems, predict environmental changes, and design new materials with unprecedented precision. These aren’t just incremental improvements—they’re paradigm shifts.

Consider the implications. In materials science, AI can now predict the properties of new compounds before they’re synthesized in a lab, slashing R&D timelines from years to months. In climate science, Google’s models are helping predict extreme weather events with greater accuracy, enabling earlier evacuations and better resource allocation. And in healthcare, AI-driven diagnostics are beginning to outperform human specialists in detecting conditions like diabetic retinopathy and certain cancers.

What sets Google apart in this space is its unique combination of computational power, data access, and interdisciplinary collaboration. Unlike startups or academic labs, Google can deploy massive AI models across global infrastructure, train them on petabytes of scientific data, and integrate them into real-world workflows. This scale is unmatched—and it’s why, despite its struggles in consumer-facing AI, Google remains a dominant force in scientific AI.

The Health Frontier: AI as a Medical Revolution

One of the most exciting—and ethically complex—areas where Google is making strides is in healthcare. The company’s AI research has already demonstrated the ability to detect diseases from medical images, predict patient outcomes, and even assist in surgical planning. But the real revolution may come from integrating these tools into everyday clinical practice.

Imagine a world where your doctor’s AI assistant can analyze your medical history, genetic data, and real-time biometric readings to recommend personalized treatments. Or where AI systems can monitor hospital patients 24/7, flagging early signs of sepsis or cardiac arrest before human staff would notice. Google is actively developing systems that do exactly this.

For example, Google Health has been working on AI models that can interpret chest X-rays with radiologist-level accuracy, helping doctors in underserved regions make faster, more reliable diagnoses. In another project, the company trained an AI to predict the onset of acute kidney injury up to 48 hours in advance—giving clinicians a critical window to intervene.

But with great power comes great responsibility. The use of AI in healthcare raises serious questions about privacy, bias, and accountability. Who is liable if an AI misdiagnoses a patient? How do we ensure these tools don’t perpetuate existing health disparities? Google has been working with regulators, ethicists, and healthcare providers to build responsible AI frameworks, but the challenges remain significant.

The Bigger Picture: Beyond the Hype

As Google takes the stage at I/O, it’s important to remember that the most impactful technologies often emerge not from the loudest announcements, but from the quiet, persistent work of researchers and engineers. While the world watches for the next flashy AI demo, the real transformation may be happening in labs, hospitals, and climate centers around the globe.

Google’s journey reflects a broader truth about innovation: leadership is not permanent. The company that once defined the AI era is now fighting to reclaim its place. But in its struggle, it may be rediscovering what made it great in the first place—a commitment to solving hard problems, not just winning races.

In the end, Google’s story at I/O 2025 may not be about who’s ahead in the coding race. It may be about how a company, faced with its own limitations, chooses to reinvent itself—not by chasing trends, but by returning to its roots: using technology to understand and improve the world. And in that mission, Google may yet find its greatest victory.

This article was curated from What to expect from Google this week via MIT Technology Review