Popular browsers are quietly reserving significant local storage for AI models, prompting new capacity and policy considerations for managed and personal PCs amid evolving on-device AI features.
Google Chrome and Microsoft Edge are quietly reserving a sizeable amount of local storage for on-device artificial intelligence features in Windows 11, creating a new capacity issue for managed and personal PCs alike. According to Google’s help documentation, Chrome may download local generative AI models in the background or when a feature first needs them, and the browser expects roughly 20 GB of free space, plus an unmetered connection and sufficient device performance.
That local model is intended to speed up selected browser tasks by processing some requests on the device rather than sending them to Google’s servers. Reports on Chrome’s storage behaviour note that the model used for Gemini Nano-based features can occupy several gigabytes, with the browser storing the files in its profile data and removing them automatically if resources become constrained. Google also allows users to disable on-device AI in Chrome’s settings, which turns off features that depend on the model.
Microsoft Edge is adopting a similar approach, but with stricter hardware and storage conditions. The browser is said to require about 20 GB of free space on the volume that holds the Edge profile, at least 5.5 GB of GPU VRAM and a non-metered connection before it will fetch the model. Microsoft also appears to remove the model automatically if free space drops below 10 GB, a safeguard designed to prevent storage exhaustion on smaller systems.
For enterprises, the practical issue is less the AI features themselves than the policy and capacity implications. Chrome administrators can govern model downloads through Chrome Enterprise controls, while Windows 11 Registry policies discussed in recent coverage can block automatic installation or reinstallation of these local models. That makes storage planning, GPU eligibility and network policy part of browser management, particularly where devices are tightly provisioned or operate on metered links.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
