The Galaxy S10 will be the first Samsung phone to include advanced artificial intelligence (AI) and machine learning (ML) features, including improved photography and voice recognition thanks to a neural network that will be built right into the processor. Various rivals like Apple and Huawei have already deployed the technology in 2017 and 2018 smartphones. Google, meanwhile, has a Pixel Visual Core chip inside its latest Pixel phones that handles photography.
Now, a report from Korean site ETNews says that the Galaxy S10’s chip will pack a dual-core neural network to handle AI and ML features such as image processing, image recognition, and speed recognition.
The chip will be manufactured on new 7nm EUV chip manufacturing technology, something Samsung announced a few weeks ago without explicitly mentioning any devices. At the time, we speculated that the Galaxy S10 chip would be a 7nm EUV processor. After all, the iPhone XS, iPhone XR, and Huawei Mate 20 Pro, which happen to be the Galaxy S10’s main rivals, all pack 7nm chips. These devices also have dedicated chip cores for neural engines that handle AI and ML features, with both Apple and Huawei designing their silicon in-house.
Samsung also makes its own mobile chips under the Exynos brand, but Samsung usually relies on a mix of Exynos and Qualcomm chips to power its flagship devices. The Galaxy S10 is expected to come in Exynos and Qualcomm versions just like its predecessors. Both chips should feature built-in neural networks.
Meanwhile, Google’s Pixel 2 and Pixel 3 feature a Google-made Pixel Visual Core processor that helps with computational photography. Other smartphones may advertise AI features of their own, but that doesn’t necessarily mean their processors have neural engines to power them.
The Galaxy S10 should be released at some point in the first half of 2019, with MWC being the likely venue for Samsung’s announcement.