With a solid manufacturing base already established, analysts see Vietnam as the ideal partner for transferring tech and innovation know-how.
Taiwanese entrepreneurs had been exploring opportunities in Vietnam long before discussions about a “China Plus One” strategy of minimizing supply chain dependency on China by diversifying sourcing emerged among Western companies. By 2013, Taiwanese investment in the country had peaked at more than US$1.7 billion, with the vast majority of capital concentrated in the manufacturing sector.
With 80% of the roughly 4,000 Taiwanese companies in Vietnam engaged in manufacturing in 2022, the sector remains the driving force of the relationship. In recent years, however, companies have started shifting their focus toward leveraging cutting-edge Taiwanese tech to aid Vietnam in industrial transformation.
“A lot of manufacturers have gone to Vietnam thinking about China in the ’90s – cheap labor and how they can use this,” says Jerry Huang, cofounder and CEO of Profet AI, a Taiwanese company that provides artificial intelligence and machine learning (ML) solutions to the manufacturing industry. “But that’s not what the Vietnamese government wants. They want to be advanced and move up the value chain.”
Profet AI and like-minded companies are establishing a “different version of the traditional model” that is helping countries such as Vietnam upgrade and achieve their development goals. “This is a role that Taiwan is uniquely able to perform,” Huang says.
Hanoi’s desire to redirect the economy away from labor-focused manufacturing toward innovative, data-driven enterprises has resulted in the eager courting of second-tier Taiwanese integrated circuit (IC) companies. “These are the ones investing,” says Huang. “They can’t get the top talent in Taiwan, but they can in Vietnam.”
This view is echoed by Matt Ryan, a Taipei-based communications consultant. “The sweatshop model of manufacturing is largely over,” says Ryan, who has worked with Profet AI. “Nobody wants to work in factories anymore. It’s about upskilling.”
There is also dissatisfaction with the trend of foreign companies moving in and out as soon as they find cheaper sources of labor elsewhere. “Vietnam, in particular, is focused on attracting full supply chains rather than just individual companies,” says Ryan. Nurturing and retaining domestic talent is another area where AI can help. “They are playing the long game and looking for sustainable growth,” he says.
Ryan highlights Vietnam’s large, educated, and perhaps most importantly, young population as an important factor in the government’s desire to shift toward innovation – a view that is supported by other industry analysts.
“Two of the main considerations for foreign companies are whether there’s a young population of workers available and their level of education,” says Kenneth Tan, managing director of Singapore-based Eastern Trade Media, which features a magazine dedicated to industrial automation in Vietnam. “Vietnam fits on both counts,” he says.
From training workers to replicating Taiwan’s core know-how in new locations, the potential benefits of AI for manufacturing are manifold. One major advantage is the predictive capabilities it offers, which can be harnessed to address production line problems. Since companies typically react to issues such as line stoppages only after they have happened, considerable capacity loss is inevitable while investigations are conducted, and solutions are implemented.
“There’s so much data in the manufacturing line, including [information on] the material, the equipment, or the environment – even the humidity and the temperature,” says Jonathan Yu, general manager of Profet AI’s Global Sales Department. “AI can help diagnose the particular defect ahead of time,” while also offering recommendations on how best to adjust settings in advance to minimize potential disruptions.
However, several practical challenges hinder AI adoption in Vietnam, including a lack of internal analytics skills and qualified data science experts. To address this, Profet AI offers automated machine learning (autoML) solutions, eliminating the need for manual coding. As this “requires different kinds of iterations and is very time-consuming,” it usually takes two to three months to create a prediction model to address a specific need, notes Yu.
Yu says that this is too slow for manufacturers needing to maintain a breakneck pace. In contrast, the Profet AI platform allows clients to upload the relevant data and achieve the same results within a couple of weeks. “As long as users have a pre-defined problem or business challenge and structural data, they can come out with a prediction model with just a few mouse clicks,” he says. “No coding or technical skills are required.”
Identifying suitable use cases is a further consideration for manufacturers hoping to implement an AI-based strategy. To assist with this, Profet AI maintains a library of industry-specific “plug-and-play” applications across a range of industries, including semiconductor, petrochemicals, and food and beverage. With 200 of these ready-to-go apps currently available, customers can browse the archive to see if there is something to fit their needs and then discuss options with the Profet AI team. These apps can also be scaled depending on clients’ requirements.
The art of governance
A component of the platform relates to AI governance – the management and monitoring of AI practices within organizations – which is itself achieved through AI solutions. “Companies in Europe and the U.S. are all talking about AI governance,” says Yu. “This has become a key challenge, so we want to preserve all the data, knowledge, and key factors, so they can be transferred to different use cases.”
For companies that are expanding to different regions, the platform’s sharing function will allow for the seamless replication of ML models in new locations. In rolling out operations to new sites, a key obstacle is human resources, says Yu. “It’s not about the equipment or the factories because if you have the budget, you can find those. The hard part of transformation is finding experienced workers.”
Noting that highly qualified Chinese employees are usually reluctant to relocate with their companies, Yu says the aim is to create a user-friendly, knowledge-sharing system that can facilitate the simple transfer of models to “companies that are going global.”
For Chin Gia-lung, a consulting partner for PricewaterhouseCoopers (PwC) in Taiwan, the role of AI in addressing environmental, social, and governance (ESG) concerns will also be crucial for many manufacturers in Southeast Asia.
“How we leverage AI to be involved in carbon credit has become a major talking point in Taiwan and Southeast Asia,” says Chin, who focuses on innovation for Taiwanese companies in Southeast Asia. He refers to the launch of the first phase of the European Union’s Carbon Border Adjustment Mechanism (CBAM) in October as an important factor. “The guidelines cover not only the Taiwanese [parent] corporations, but also their overseas entities,” he says.
To prepare for such developments, PwC is cooperating with Microsoft and OpenAI, among others, on a program that focuses specifically on AI solutions for carbon offsetting. “Carbon emissions are a critical issue in Vietnam anyway, so this is something we’re focusing on,” says Chin.
He highlights the cooperation between universities in Taiwan and Vietnam to train a qualified Vietnamese workforce. These efforts are sponsored by Taiwanese NGOs and corporations that have a vested interest in fostering a new generation of workers who are equipped with the requisite skill sets. “The Taiwanese corporations know what type of resources they need, so they help tailor specific programs for universities over there,” says Chin.
Many of these businesses are the IC companies mentioned by Huang. “The Vietnamese authorities are rolling out the red carpet for these companies, who are in turn investing in education.”
Educational initiatives are also supported by companies such as Profet AI. “We’re building alliances with different universities,” says Yu, who mentions internships for students at National Taiwan University and National Taiwan University of Science and Technology as cases in point. “There’s also cooperation with AI Academy, which is the largest [AI-focused learning institution] in Taiwan. We’re putting these resources together and bringing them to Vietnam.”
Some observers question the feasibility of moving too quickly. “It would probably be very difficult for Vietnam to immediately replace all the labor-intensive industries with automation and AI,” says Tan. He cites Vietnam’s “fairly mild Covid situation” as having influenced this. “Unlike some countries, there wasn’t this big shock over how to do production when people couldn’t get into factories,” he says. “Because there wasn’t that particular motivation for lesser reliance on manpower, the lesson probably hasn’t sunk in.”
There may also be a need to first consolidate data for many manufacturers. “If you haven’t collected sufficient data within a manufacturing environment, then you can’t do much with AI,” says Tan. “So, one needs to look at which companies and industries will truly find this useful.”
However, Yu believes this is not a major issue. “Many of our customers worry that they need a lot of data and that it needs to be well-integrated into a platform or database,” he says. “But even with just Excel or [comma-separated values] CSV files, there’s a lot of experience and insight available.”
There are still fears that digital transformation may leave unskilled manual workers behind. Yu believes these concerns miss a crucial point about digital transformation. “Of course, automation will replace some low-level workers, but AI is a decision-supporting system,” he says. “It’s like the saying goes – people won’t be replaced by AI, but by people who know how to use AI. That’s why we make our solutions easy, so that even employees with no coding skills can gear up.”
He emphasizes the triangular structure of traditional enterprises, with high-level executives at the top, followed by middle management, then workers “just following SOP” at the bottom. The goal, he says, is to transform this structure into a diamond shape. “The middle level in manufacturing works more on data analysis and improvement,” says Yu. “By training low-level workers in these skills, we can move them to the middle.”
Regardless of the pace of the transition and the potential discomfort it may cause, the AI revolution in manufacturing is seen as inexorable by most, with Taiwan ideally placed to take advantage.
“In the long-term, it’s a must-happen trend,” says Chin. “And there’s a vantage point for Taiwanese AI companies in Vietnam because they already have years of experience and an established manufacturing base there.”