Chinese networking giant Huawei has unveiled its Kirin 970 chipset with built-in artificial intelligence (AI), calling it the "future of smartphones".
The chipset has a dedicated neural processing unit (NPU), and was built using a 10nm advanced process with 5.5 billion transistors contained in an area of just one square centimetre.
The Kirin 970 mobile AI computing platform is powered by an 8-core CPU and a 12-core GPU, which Huawei said delivers 25x greater performance and 50x greater efficiency compared to a quad-core Cortex-A73 CPU cluster.
During an image-recognition test for benchmarking purposes, Huawei said the chipset's performance saw it process 2,000 images per minute.
Huawei has additionally provided the Kirin 970 as an open platform for mobile AI developers and partners.
According to Huawei Consumer Business Group CEO Richard Yu, the company's mobile AI is made up of a combination of on-device AI and cloud AI.
"Huawei is committed to developing smart devices into intelligent devices by building end-to-end capabilities that support coordinated development of chips, devices, and the cloud," Yu said.
"The ultimate goal is to provide a significantly better user experience. The Kirin 970 is the first in a series of new advances that will bring powerful AI features to our devices and take them beyond the competition."
Limitations in cloud AI necessitated improvements across latency, stability, and privacy, Huawei said, with on-device AI providing this as well as adding sensor data to the offering.
"Sensors produce a large amount of real-time, scenario-specific, and personalised data. Supported by strong chip processing capabilities, devices will become more cognitive of user needs, providing truly personalised and readily accessible services," Huawei said.
The Kirin 970 announcement follows Huawei last month saying AI would play a critical role in driving smartphone innovation.
Bruce Lee, global VP of handsets business, told ZDNet in August that Huawei is focusing some of its research and development efforts on AI hardware and software, particularly within handsets themselves, rather than sending sensitive data back and forth between device and cloud.
"We hope to use AI in our phones to have more learning capabilities ... together with big data, we will be able to understand consumer habits and better incorporate voice and image capabilities into the phone," Lee told ZDNet.
"We can then have faster responses because we don't need to upload data from the device into the cloud, do the computing, and send it back into the device. And when we do the computing on the local device, we can also safeguard user privacy since we don't need to upload data into the server."
Huawei's 2016 annual report [PDF] similarly discussed a time of "+Intelligence" wherein all devices, processes, and people would be supported by AI.
"Building intelligence into our devices, networks, and industries will open up new worlds," Huawei said.
"AI will disrupt the user experience, but before it can do so, we will need a quantum leap in the functionality of our smart devices, chipsets, and cloud services.
"Artificial intelligence will place heavy demands on computing performance, energy efficiency, and device-cloud synergy. Meeting these demands and creating a better intelligent experience will take a synthesis of capabilities across both chipsets and the cloud."
Huawei first predicted the advent of the "superphone" almost two years ago, saying it would be developed by 2020 and take advantage of advancements in AI, big data, and cloud computing.
"The intelligence of the superphone will continue to evolve and develop itself into digital intelligence, capable of empowering us with interactions with the world," Shao Yang, president of Strategy Marketing at Huawei Consumer Business Group, said in November 2015.
"Through evolution and adaptation, the superphone will be more intelligent, enhancing and even transforming our perceptions, enabling humans to go further than ever before."