Since both HiSilicon's Kirin 980 and Apple's A12 Bionic application processors come with AI (artificial intelligence) accelerators, the penetration rate of related AI-based accelerating solutions in smartphone production is expected to surpass 30% in the fourth quarter of 2018, according to Digitimes Research.
 
While the Kirin 980 is packed with two Cambrian 1H neural network processing units (NPUs) to boost its AI computing performance, the A12 Bionic chip optimize its computing ability by integrating iOS 12 and ARKit 2 to enable the new iPhone XS, iPhone XS Max and iPhone RX, to enhance their computing performance.
 
If divided by hardware acceleration driven by NPUs or software acceleration synced up by GPU, DSP and other units with computing resources available within APs, it is expected that the adoption of NPU-based accelerating solutions will be higher than those software-based solutions in the fourth quarter of 2018, Digitimes Research estimates.
 
While facial recognition and image processing are currently the most sought after AI tasks to be implemented on smartphones, the proliferating AR applications will challenge the capabilities of next-generation APs. As a result, smartphone vendors and application developing communities have to come out with differentiated AI applications to continue to attract consumers, while fending off rivals.