The Rise of AI in the Manufacturing Sector

By: Marcus Chu

Investment Analyst

Artificial intelligence (AI), an enabler in many manufacturing verticals, has been growing rapidly with more and more use cases in recent years, resulting in a higher automation penetration rate in manufacturing supply chains.

AI has been adopted in many vision-related applications such as defects detection and facial recognition as it is able to process complex images and improve detection accuracy compared to simple rule-based vision functions. AI is also deployed in data analytics, model prediction, robot path guiding, etc. With AI-powered solutions, industry pain points have been greatly addressed. For instance, robotics with vision processing ability has been deployed in many new applications and challenging tasks such as grabbing and holding heavy items, or welding and cutting metal sheets.

In the manufacturing sector, quality sensors are key to training AI models as they can capture necessary operating data, including but not limited to visual images, operating data, and non-visual measurements such as flow speed, vibration, noise, etc. On top of a solid AI software solution, companies have to pick compatible and competitive hardware to best cater to customers’ demands.

To deploy AI solutions on their supply chains, manufacturers can either do it by themselves or rely on third-party solution providers. It is usually challenging for corporates to build their own expertise due to high capital expenditure and operating expenses to fund a team of software engineers. While third-party solution providers bring with their specialty in software solutions and good connections in hardware parts. They also enjoy lower costs, thanks to economies of scale, and are in a better position to suggest changes to existing factory designs. The main concern with third-party solutions is potential leakages of data and technology to competitors in the industry.

It takes time and effort for AI companies to develop enterprise solutions for different industries. The first use case often comes at a high cost. It also takes time for manufacturers to build commitment and embrace potential disruptions to their existing production lines. But as there are more of these use cases and AI models/assets developed, we expect the speed of new deployment to accelerate. Since the AI data sets are reusable, it will translate to lower project costs and better project margins. We also note potential replication in many other similar factories leveraging the scalability and economies of scale.

In China, two AI companies, namely SenseTime and AInnovation went for IPO in Hong Kong recently. SenseTime is the largest AI software company in China. A majority of its revenue, according to the prospectus, comes from smart business and smart city solutions. AInnovation is specialized in intelligent manufacturing. One key milestone project is to restructure and automate the transportation of high-temperature molten iron in steel factories. Both companies have received decent market attention and we expect more AI companies to file for IPO in the future.