Remember when a web browser was just a window to the internet? A lightweight, snappy application whose only job was to render a page and get out of your way? In 2026, those days are definitively over.
Google Chrome and Microsoft Edge—the two undisputed titans of the browser market—have decided your RAM is their RAM. The culprit? An aggressive, uninvited, and increasingly heavy integration of Artificial Intelligence.
Google recently released a new “Prompt API” with chrome(and every software that uses it, like Edge) which means it now ships with Gemini Nano. That extra 2-4GB of RAM those browsers are using now? An ENTIRE LLM shoved into your memory.
Good thing RAM is cheap now right?
The Consequences
The impact is such a stupid decision is clear:
- Slower performance: More RAM usage means more swapping to disk, which is up to 10x slower.
- Security risk: The AI has access to a lot of information about you and your machine
- Frequent crashes and reloads: Tabs and apps are more likely to be terminated to free up memory.
- Reduced battery life: Constant background processes drain power faster.
- Shorter device lifespan: Users are forced to upgrade hardware sooner than necessary.
There are alternatives
Firefox, browsers who use firefox, and safari don’t have such stupid API and at least mozilla says they don’t plan to implement it any time soon:
“We continue to oppose this API, and feel it has severe negative consequences to the interoperability, updatability, and neutrality of the web platform,”
– Archibald, Mozilla web developer relations lead
So if you wish to save some RAM, CPU and GPU, avoid chrome based browsers(Google Chrome, Edge, Opera, etc).
The Illusion of Benchmarks
Also, if you look at synthetic speed benchmarks (like Speedometer or JetStream), Chrome and Edge still often claim the top spots. But these lab tests are lying to you.
Benchmarks measure raw JavaScript execution in a vacuum. They do not measure interaction delays (INP), memory leaks, or the sluggishness you feel when you have 20 tabs open while a local AI model is simultaneously trying to index the text on your screen. In the real world, “fastest in the lab” translates to a browser that spikes your RAM usage to 80% just to keep its background AI services idling.
