Machine vision and automative applications are propelling the demand for AI chips to new heights...
Big, complex data sets, a growing list of commercial applications advanced by impatient consumers and widespread adoption of deep learning and neural networks are seen as accelerating the AI chips set at a roughly 40-percent annual growth rate, according to a new outlook.
The AI technology forecast released this week by MarketsandMarkets pegs the global chip set sector at a vibrant $57.8 billion by 2026, Computer vision applications for industrial and automotive applications are expected to register the highest annual growth rate as more machines become at least semi-autonomous.
Meanwhile, the market for graphics processors that serve as the work horses in datacenters for training machine learning models is seen expanding at the fastest annual rate, the analyst said. That expansion will likely be fueled by GPU leader Nvidia Corp.’s acquisition of chip IP vendor Arm Ltd.
“Arm expands Nvidia’s developer reach from 2 million to over 15 million,” Nvidia CEO Jensen Huang noted in announcing the blockbuster acquisition. “The deal will also bring tremendous benefits to customers who will offer Arm customer’s access to Nvidia’s AI and GPU IP.”
AI applications such as machine learning also will drive adoption of slower but more power-efficient FPGAs, the analyst said, while x86 CPUs continue to play a supporting role in accelerating specific deep learning workloads that currently dominate enterprise datacenters.
Much of the demand for AI chip sets over the next five years will come from the Asia Pacific region, according to the forecast, especially automotive and industrial applications being rolled out in China, South Korean and Japan. The key regional drivers include declining AI hardware costs, AI IC performance improvements and smartphone-obsessed consumers clamoring for instantaneous service.
Hence, advancing AI hardware with improved latency and real-time responsiveness will spearhead the regional explosion of automated services.
A familiar roster of makers of AI chips is expected to dominate the market over the next five years. Along with Nvidia and its Arm unit, Intel Corp., Samsung Electronics, AMD, IBM with its Power architecture along with FPGA specialist Xilinx head the list.
Meanwhile, hyper-scalers such as Amazon, Google and Microsoft are seen playing a larger role in the AI software stack.
The survey also lists embattled Huawei Technologies among the AI chip leaders, but overlooks China’s other key AI players, including Baidu and Tencent. Meanwhile, China cloud giant Alibaba unveiled its first AI chip, dubbed Hanguang 800, last December.
Another emerging player in the AI chip sector is Hailo, the Israeli startup that pitches its AI processors for edge devices like unattended sensors. The Tel Aviv-based startup completed a $60 million funding round earlier this year that will be used to ramp up AI chip production.
In addition, AI (chipsets) market in APAC projected to grow at the highest CAGR from 2020 to 2026. This growth can be attributed to the adoption of AI services in end-user industries such as manufacturing, healthcare, and automotive in countries such as Japan, China, Australia, and South Korea. Additionally, declining AI hardware costs and increasing demand to improve consumer services are supplementing the growth of the AI (chipsets) market in APAC.