MacBook Air M2 - High CPU Usage and Temperature Spikes with Google AI Studio

Hi Google AI Studio Team and fellow users,

I’m experiencing intermittent CPU usage spikes and temporary temperature increases (up to 80°C, then quickly cooling) when using Google AI Studio on my MacBook Air M2 (passively cooled, running macOS 15.3.1). This happens even during seemingly “low usage” periods.

Through investigation with Activity Monitor, I’ve isolated the issue to the Google AI Studio web application itself, not a background process or connected devices. The spikes are much more pronounced when the AI Studio tab is in the foreground (visible) and correlate strongly with the real-time timer display. Switching to another tab significantly reduces CPU usage and temperature, even while the model continues generating a response.

My analysis:

  • The issue is likely due to client-side JavaScript processing within the AI Studio web app, specifically the frequent updates of the timer display and the handling of the model’s responses.
  • Modern browsers aggressively throttle JavaScript in background tabs, explaining the rapid temperature drop when switching tabs.
  • The MacBook Air M2’s passive cooling handles the heat well after the CPU load decreases, but can’t prevent the initial spikes.

This behavior seems to be “normal” for this type of web application, but it’s worth noting for users with passively-cooled devices, especially in warmer climates (I’m in India, with ambient temperatures around 35°C).

Workarounds I’m using:

  • Minimize Foreground Time: I switch to another tab while waiting for long responses and check back periodically.
  • Close Unnecessary Tabs: I close other browser tabs to free up resources.
  • Cooling Pad: I use a cooling pad.

Suggestion for the AI Studio Team:

Could the timer update frequency be optimized, or perhaps offer an option to disable the real-time timer display to reduce CPU load, especially for users on passively-cooled devices?

Has anyone else experienced this? Any other tips? Thanks!

Me too. The link shows a high CPU usage screenshot.

You can try: