Samsung to double Gemini-powered devices to 800 million this year

Samsung Electronics plans to double the number of its mobile devices with AI features powered by Google’s Gemini to 800 million units in 2026.

The South Korean company had rolled out Gemini-powered AI features to about 400 million smartphones and tablets by the end of last year.

Samsung, which is the largest supporter of Google’s Android platform, said it aims to bring AI to all products and services as quickly as possible.

The company is seeking to strengthen its position in the smartphone market against Apple and Chinese competitors, while also expanding AI integration in TVs and home appliances.

🔗 Source: Reuters

🧠 Food for thought

Implications, context, and why it matters.

Samsung’s 800 million AI device goal faces hardware and market hurdles

Samsung plans to double Gemini-powered devices to 800 million by 2026, but the smartphone market could shrink then due to memory chip shortages. Rising chip prices are squeezing margins, and the company has not ruled out higher phone prices 1. As of October 2024, Gemini Nano (a lightweight, on-device version of Google’s Gemini AI model) runs on six Galaxy models 1.
Gemini Nano needs AI accelerators, which are specialized processors for machine learning tasks, in chips like Samsung’s Exynos 2400e and Qualcomm’s Snapdragon 8 Gen 3 1 2. Many older phones lack on-device support, so Samsung leans on cloud-based processing (running AI features on remote servers) 2. Cloud use adds latency, a noticeable delay, and it raises privacy questions since features may feed model training that improves AI models with user data 3 2. Hitting 800 million needs wider on-device support in mid-range phones or broader cloud use.

Third-party developers have limited tools to build on Samsung’s AI base

Developer tools lag. Samsung pulled its Neural Software Development Kit (SDK) 4. There is no Samsung-run public program like Google’s AI Edge Software Development Kit (SDK) that exposes Gemini Nano, and Google’s AI Edge SDK is experimental and for development only 5. Google granted early access to Adobe and Grammarly 2.
For Software as a Service (SaaS) vendors and app makers, watch Samsung’s portal for SDK or toolchain releases and distribution programs. Infrastructure exists, based on past Exynos Neural Processing Unit (NPU) toolchains 6, though timing and Application Programming Interface (API) access remain unclear.