Most discussions about artificial intelligence take place in a world of flagship smartphones and fast internet. But over a billion people use devices with 2 GB of RAM or less — predominantly in Africa, South Asia, and Latin America, where Android controls over 85% of the market. It was for them that Google launched Gemini Go on June 3.
What changed technically
Gemini Go is not a separate application, but a component within Google Search. It replaces Assistant Go and is available on devices running Android 13 Go edition with 2 GB of RAM or more. Activation is through a long press of the Home button or power button.
Functionally, the assistant can make calls and send messages, check the time while traveling, set alarms and calendar events, play music by mood, and analyze photos and documents uploaded by the user. This is significantly more than Assistant Go could do — however, Google has not yet shown real examples of how these functions work under conditions of a weak processor and limited memory.
«Gemini Go is a simplified version of Gemini, designed to help you stay connected and accomplish tasks even on devices with smaller amounts of memory»
Anish Kotthapalli, Community Manager, Gemini Apps Team, Google
Why this matters beyond tech news
Android Go has existed since 2017 precisely as a response to market realities where most of the population cannot afford a flagship phone. In 2025, approximately 4.2 billion smartphones worldwide run Android, and the majority of new user growth is concentrated in the budget device segment. Until now, these people have been largely cut off from the wave of generative AI — not due to lack of interest, but due to hardware limitations.
Digital inequality researchers have been documenting what is called an «AI gap» for several years: advanced models become available first on expensive devices and only later — in simplified form — on budget ones. Microsoft, in its report on global AI adoption at the beginning of 2026, acknowledged that the level of generative AI adoption in low-income countries remains significantly lower than in developed ones. Gemini Go is a concrete step toward closing this gap, though it preserves significant functional asymmetry.
What remains unclear
- Which specific features of full Gemini are missing — Google did not publish a comparative table of capabilities.
- Language support — how fully Gemini Go works in languages common in target markets: Swahili, Hindi, Bengali, Hausa.
- Quality of offline operation — in regions with unstable internet this is critical, but Google did not comment on this.
- Privacy — what rules apply to the processing of voice queries and uploaded documents on budget devices.
As Chrome Unboxed notes, «this hardware requirement has traditionally been a barrier» — and now it is partially lifted. But only partially: Gemini Go still does not work on devices with less than 2 GB of RAM, and a significant portion of Android Go smartphones sold in Africa and Asia come with exactly 1 GB.
If Google, over the next year, expands support to devices with 1 GB of RAM and adds full functionality without stable connectivity — Gemini Go will become a real tool for reducing digital inequality. If not, it will remain a marketing move with narrow reach precisely where reach matters most.
