Huangshan-2 to be mass-produced in the fourth quarter of 2020, and wearable products equipped with the chip will be available next year.
On June 16, during its first AI Innovation Conference in Hefei, Huami Technology (HMI:NYSE) released five AI-based biological data engines, an optical sensor and a new generation of AI chips customized for wearables – Huangshang-2.
Huami was incubated by Xiaomi Corporation (01810:HKEX) and is publicly considered one of the key actors in the consumer electronics giant's ecosystem. As the R&D and hardware provider for various types of Xiaomi devices, Huami has collected enormous revenue from Xiaomi’s purchase orders. Though the revenue generated from Xiaomi's orders went down from 92% in 2016 to 72% in 2019, Huami's downstream structure is still imbalanced.
Being the second-largest shareholder of Huami, Xiaomi continued its mutually beneficial cooperation in the recently published Mi Band 5, which embedded the Huami-empowered health assessment system Huami-PAI.
As the core part of wearable devices, the Huangshan-2 can help take full advantage of five AI biological data engines, including Huami-PAI.
Firstly, it has two advantages: high computing efficiency and low power consumption. Compared with the ARM Cortex-M4 architecture processor commonly used in wearable devices, the overall computing efficiency has been increased by 38%.
Secondly, the performance of local AI data calculation is improved and recognition speed accelerated. For instance, Huangshan-2’s Atrial Fibrillation recognition speed is seven times that of Huangshan-1 and 26 times that of other software algorithms on the market.
Finally, in theory, the overall power consumption of Huangshan-2 may be reduced by 50%, improving the endurance capability of wearable devices.