Samsung unveils 8nm Exynos 9820 chipset with a dedicated NPU for on-device AI and 8K video recording, likely to be used in next year's Galaxy S10
Samsung is betting on a dedicated NPU to introduce a host of AI-enabled features in its upcoming flagship. — Huawei was the first to roll …
Context & Ripple Effects
Samsung's last flagship chip, the Exynos 9810, teased AI-driven face detection but left the work to general-purpose cores; the 9820 changes that by adding a dedicated neural processing unit. The description notes Huawei was first to roll out an NPU in a phone chipset, so this is Samsung following rather than leading.
The timing matters for the Galaxy S9 generation's successor: the 9820 is positioned as the silicon inside next year's Galaxy S10, making on-device AI a headline feature of Samsung's 2019 flagship rather than a background spec.
First-order effects
- The Galaxy S10 ships with a chipset built for on-device AI tasks like scene recognition and 8K video capture, letting Samsung market AI features that run locally instead of in the cloud.
- Samsung closes the silicon gap with Huawei, whose earlier NPU rollout had made it the first mover in dedicated phone-chip AI hardware.
Second-order effects
- Rival flagship chipmakers now face pressure to match the dedicated-NPU design point, since camera and AI features are becoming the visible differentiator between flagships rather than raw CPU clock speeds.
- Samsung's own roadmap locks in behind the NPU bet: the Exynos 9825 carries the same architecture forward to 7nm for the Note 10, so suppliers and software teams building around the 9820's neural engine get a multi-generation platform.
Third-order effects
- If the pattern holds, dedicated neural engines become table stakes in every flagship SoC — the later Exynos 990 doubling AI-specific performance and the 3nm Exynos 2500 pushing NPU throughput further show the investment compounding across generations.
- On-device inference shifts the phone industry's competitive axis from cloud services toward silicon-level AI capability, favoring vertically integrated players like Samsung that control both the chip and the handset.
The trend: Flagship mobile chipsets are standardizing on dedicated NPUs, turning on-device AI from a differentiator into baseline silicon design.