High labor costs and a quicker time-to-market prompts US manufacturers to be more aggressive with AI adoption.
US-based industrial manufacturers are slightly ahead of their Chinese counterparts in integrating artificial intelligence (AI) capabilities into their operations, even though both countries have a near-equal number of installed bases of AI-enabled devices, according to global tech market advisory firm ABI Research.
Both countries are incentivized to develop approaches that encourage AI adoption in industrial manufacturing, but “the US has managed to achieve more momentum,” according to Lian Jye Su, principal analyst. This is due to “acute challenges in manpower and rising cost of materials. So they need to adopt AI to make sure they can overcome these challenges.”
In China, manufacturers can still rely on “abundant human resources and government tax incentives to compete in the international market, keeping their margin healthy,” Su added.
The expensive US labor force has driven the industrial sector to enhance production efficiency and lower operational costs. This is prompting the major cloud service providers, smart manufacturing platform vendors, pure-play industrial AI platform and service providers, edge industrial AI gateway, and server and chipset vendors to partner to bring AI into industrial manufacturing, he explained.
The partnerships are enabling an “end-to-end, cloud-to-edge solution to enhance operational efficiency, reduce bottlenecks and optimize resource consumption in factories,” he said.
Industrial AI use cases
AI is predominantly used in two areas: Overall efficiency enhancement and machine vision for inspection, production, and surveillance, according to Su.
“For example, Instrumental is using images of printed circuit board to train their AI model in order to identify faults and low quality production on the smartphone production line,” he said. Other companies like Uptake and SparkCognition use historical machine data, environment data, and employee logs to construct predictive maintenance models, allowing manufacturers to get notified of their machine breakdown well ahead of schedule, he says.
Deep learning is becoming popular for use in image recognition. “Future machine vision use cases in industrial manufacturing may be relying on deep learning more and more,” Su noted.
The US industrial edge
Since 2015, the Chinese government has encouraged provincial and local government, public agencies, and state-owned enterprises to adopt AI in industrial manufacturing. Chinese cloud AI conglomerates such as Alibaba, Baidu, Huawei, and Tencent have all made AI in smart manufacturing a key priority in their business strategy, according to Su.
But while these companies “are pushing their AI solution heavily … end users are less receptive to AI,” he said.
None of the cloud companies have a background in manufacturing, but the difference is the US cloud providers “have a large number of industrial AI startups to champion their venture into industrial manufacturing,” Su said. “This is severely lacking in China,” due in part to the fact that manufacturing processes in China are still very labor intensive.
At the same time, manufacturing plants in China are still in the early stage of industrial Internet of Things (IoT) adoption in terms of linking their production equipment and deploying smart devices. Vendor relationships are also more complex in China, and vendors are typically not that familiar with AI, he said.
High labor costs and a quicker time-to-market have prompted US manufacturers to be more aggressive in adoption of industrial AI solutions. “This has paved the way for development of pure-play AI players in the US, propelling the US to be the global leader in industrial AI solutions,” Su said.
While conducting his research, Su said he had assumed momentum for adopting industrial AI would have been as equally strong in China as in the US, given that AI investment in China has been growing rapidly in recent years. However, it came as a surprise that “AI players in China are stronger in consumer and enterprise space. This also shows that implementing AI in industrial and manufacturing is a much [more] challenging proposition.”
Predictions for both countries
He believes the potential for AI to replace jobs is real, but stressed that “AI is only good at performing narrow, or specific tasks. That’s why both economies welcome it, particularly in the manufacturing industry. Workers are less and less willing to work in hazardous environments handling repetitive tasks.”
Su expects the US will continue to dominate in industrial AI. However, he adds that “China is, and always has been the most aggressive when it comes to implementation and large scale deployment. While China is definitely lagging in current industrial AI adoption, it can quickly ramp up its speed of adoption once the right market opportunity emerges.”
To stay competitive, Su recommends that US manufacturers focus on well-defined tasks, which can be replicated with the help of vendors or system integrators.
“Once these tasks have been addressed by AI, the manufacturers can then consider expanding [their] use of AI to more complex use cases and improve overall efficiency enhancement,” he said.