Dedicated processors will greatly expand the use of AI in smartphones, and voice assistants are among the first beneficiaries of AI processing "on the device.
Smartphones have been leveraging the power of artificial intelligence (AI) for some time.So far, however, the processing has been done in the cloud, or on a variety of computer chips distributed on the device, such as cpus, gpus, and DSPS.
As AI becomes part of the mobile experience, SoC vendors are improving the machine learning capabilities of their chips by integrating dedicated AI processing kernels into their designs.The main benefits are higher AI processing performance and lower power consumption.However, this has to be balanced against the actual needs of ai processing, which until recently has been limited.
At present, AI is mainly used in camera-related applications (facial recognition, image enhancement, etc.).However, Counterpoint Research expects AI to be widely used in a wide range of other applications over the next two years.
"We see voice assistants as one of the first applications to benefit from device-based processing," says Gareth Owen, deputy Research director at Counterpoint Research."Today, most speech processing in smartphones is cloud-based.However, the voice assistant will be able to process commands faster and process faster responses on the device."" it also addresses privacy issues," "he added.
Apple and huawei were the first oems to add dedicated AI processors to their flagship phones, the A11 and Kirin 970 chips.Two years later, almost every other SoC vendor followed suit.Qualcomm, for example, is the first to offer the AI Tensor Accelerator in Snapdragon 855's Hexagon DSP.
By 2020, sales of artificial smartphones will increase from 190 million in 2018 to 1.25 billion, accounting for more than three-quarters of smartphone shipments that year, predicts Counterpoint Research.
A key driver behind the forecast is the expected launch of several Android smartphone models using qualcomm's 855 SoC in 2019 and beyond.Counterpoint expects that the dedicated ai core, which USES the new P90 SoC from Taiwanese vendor mediatek, will also be incorporated into some mid-range smartphones during 2019.
While smartphone users may not see AI immediately, smartphones with dedicated AI hardware will clearly have the edge as the market for machine learning grows and more applications emerge.
"Industry-wide adoption of AI chips in smartphones is inevitable as qualcomm enters the fray and for the first time incorporates hardware-based AI acceleration technology in its latest Snapdragon SoC," said Peter Richardson, Research director at Counterpoint Research."With qualcomm keen to extend machine learning capabilities to its low-cost chips, and with other mid-range smartphone AI socs from other manufacturers, this trend will rapidly expand to mid-range and even low-end smartphones," he added.