Google Cloud goes all-in on Gemini

Google just went all-in on embedding its flagship AI across its entire enterprise stack.

At Cloud Next, the company unveiled Gemini for Google Cloud, infusing generative AI into developer environments, infrastructure operations, cybersecurity, and data analytics. Leading the charge is Gemini Code Assist—the successor to Duet AI—now supercharged by the massive 1-million-token context window of Gemini 1.5 Pro.

The push comes as Google Cloud hits a $36 billion annual revenue run rate, five times its pace from five years ago. Google also noted that nearly 90% of generative AI unicorns are now its cloud customers.

An enterprise-wide AI overhaul

Gemini Code Assist is aiming straight for developers' day-to-day workflows. Available in editors like VS Code and JetBrains, it connects to private codebases across GitHub, GitLab, Bitbucket, and on-premises setups. Currently in private preview, its 1-million-token full codebase awareness allows it to track cross-file dependencies, manage complex refactoring, and conduct deep code reviews. Partner Quantiphi reported internal efficiency gains topping 30%.

Beyond coding, Gemini Cloud Assist acts as an in-console AI architect, generating architecture setups from natural language and troubleshooting system outages.

In security, Gemini integrates with Chronicle and Mandiant threat data to parse malicious activity and summarize alerts. For data teams, Gemini in BigQuery introduces a visual data canvas, turning natural language prompts directly into SQL, Python, and interactive charts.

Simultaneously, Gemini 1.5 Pro entered public preview across 180+ countries via the Gemini API and Vertex AI, introducing native audio understanding, JSON mode, and system instructions.

Why it matters

We think Google's sharpest edge here isn't just another conversational assistant; it is operationalizing the 1-million-token context window where it actually solves massive pain points. Legacy codebases and complex multi-repo architectures routinely break standard AI coding assistants that only "see" a file or two at a time. Shoving an entire codebase or mountains of system logs directly into context genuinely changes how teams handle refactoring and incident triage.

For enterprise teams and engineers, it is well worth testing Gemini Code Assist if you deal with sprawling, tangled microservices where context switching is your biggest time sink.

This is Google's clearest counterpunch yet to Microsoft's GitHub Copilot and AWS. While Microsoft locked down developer mindshare early, Google is pitching a unified, full-stack story spanning from underlying TPU v5p chips to BigQuery and security ops. The lingering hurdle will be convincing privacy-sensitive enterprises to comfortably pipe their most confidential code repositories through Google's cloud.