Google Ships VS Code Extension for Remote Workbench Execution

On July 1, 2026, Google officially launched the Google Cloud Workbench Notebooks extension for Visual Studio Code. The release allows data scientists and machine learning engineers to connect their local editor directly to scalable Jupyter environments running on Google Cloud infrastructure. For developers handling large-scale training runs, this architecture bridges the gap between local workflow productivity and remote GPU or TPU compute.

Historically, managing cloud-based notebooks required switching between a local integrated development environment and a browser-based JupyterLab interface. This extension maps the remote execution environment directly into the local editor.

Remote Kernel Execution

The extension bypasses the standard browser UI by mapping a remote Gemini Enterprise Agent Platform Workbench instance (formerly Vertex AI Workbench) as a local Jupyter kernel. Developers open a .ipynb file in VS Code and navigate to the kernel selection menu. By choosing the Google Cloud Workbench option, the execution runs on high-performance cloud hardware while the developer maintains their local themes, keyboard shortcuts, and installed extensions.

This architecture ensures that large data payloads and heavy compute cycles stay in the cloud network. Your local machine only handles the interface state, preventing memory bottlenecks when iterating on large datasets.

Ecosystem and Telemetry Integration

Google open-sourced the extension to encourage community contributions and auditability. The release coincides with a broader restructuring of Google’s developer tools under the Antigravity platform.

Alongside this extension, Google updated its Cloud Code extension and introduced the Antigravity CLI to replace legacy Gemini Code Assist tiers. For enterprise teams, the integration natively supports OpenTelemetry via the AI Telemetry Collector released in late June 2026. This allows VS Code users to monitor standardized TPU metrics and identify silent failures during multi-node training directly from their connected environment.

Setup and Specifications

Routing your IDE to cloud infrastructure requires specific dependencies. The package is available in the Visual Studio Marketplace and Open VSX under the official namespace.

Component	Requirement
Package Name	GoogleCloudTools.workbench-notebooks
Prerequisite Extension	Jupyter extension for VS Code
Cloud Environment	Active Google Cloud Project
Platform Requirement	Gemini Enterprise Agent Platform enabled

Authentication uses OAuth to link the local environment to the target GCP project. Once authenticated, the extension populates a list of active Workbench instances available for kernel execution.

If your team relies on provisioning cloud GPUs for heavy machine learning tasks, this extension centralizes the workflow. Standardize your remote Workbench instances in GCP, install the extension across your team, and execute cloud-scale code without leaving the local editor.