This article first appeared in Forum, The Edge Malaysia Weekly on November 4, 2024 - November 10, 2024

Recently, there was a forum held in Penang that brought together foreign and local chip manufacturers and learning institutions to discuss how to meet the goal of training and upskilling 60,000 engineers as part of the National Semiconductor Strategy (NSS). According to Malaysia Semiconductor Industry Association (MSIA) chairman Wong Siew Hai, companies such as Oppstar, Skyechip, 

Infinecs, Advanced Micro Devices (AMD) and Intel already employ more than 7,000 engineers from the current pool of about 90,000, but there is concern that the number of engineers may not be sufficient to help meet MSIA’s goal of achieving RM1.2 trillion in chip production by 2030.

The increasing need for engineers is driven by the growing demand for chips for generative AI (artificial intelligence). Mckinsey estimated that following the release of ChatGPT in November 2022, generative AI could add US$4.4 trillion to the global economy annually. This figure surpasses Japan’s entire GDP of US$4.2 trillion in 2023. Generative AI operates through large language models (LLMs), which learn patterns, grammar and knowledge from massive datasets that enable today’s powerful chatbots and are capable of creating new content. The LLM market reached global sales of US$4.1 billion in 2023 and is expected to grow significantly to almost US$54 billion by 2032.

Generative AI relies on advanced chips that handle massive amounts of data and complex neural networks that are trained to identify patterns and make inferences. These chips are integrated into various specialised hardware including graphic processing units (GPU), tensor processing units (TPU), application-specific integrated circuits (ASIC) and field-programmable gate arrays (FGPA) for AI computing, memory, storage and networking purposes. To scale AI capabilities, more skilled engineers are needed to develop more powerful hardware. A Boston Consulting Group survey reveals that the US — being the largest AI producer — alone needs about 400,000 new engineers.

The supply of engineers remains a critical gap despite significant layoffs at large tech companies. The long-term outlook remains positive due to strong investment and increasing job postings in generative AI and data centres. Designing chips for generative AI requires group knowledge and skills from hundreds of engineers, with the cost of developing a 5nm chip reaching US$540 million. The design stage adds the most value — over 50% — compared to wafer fabrication (19%) and manufacturing equipment (12%) as well as smaller contributions from assembly, test and packaging (6%), materials (5%), electronic design automation (EDA) and core intellectual property (IP) (3%).

While developing advanced chips requires skilled engineers of different roles, chipmakers are facing worldwide challenges of too-few graduates, an ageing workforce and legacy workers unwilling to learn new skills unless the pay prospects are much better. When chipmakers are competing for an overstretched talent pool, upskilling existing engineers becomes the alternative solution. According to Bain Tech Talent Survey 2023, generative AI is being reinvested into the work tasks of engineers to boost productivity. Around 57% of engineering teams are already using AI such as coding assistants to design chip functions, with many reporting that AI helps complete about 20% of tasks faster while retaining quality.

For instance, Nvidia’s ChipNeMo helps engineers generate scripts for EDA tools that enable engineers to identify circuit logic using natural language commands. These tools — a chatbot, a code generator and an analysis tool — combined with human engineers’ assistance help scale the AI chip’s capabilities. While generative AI can be a key part of an upskilling strategy for engineers, new skills could reduce the size of engineering teams. Mckinsey identifies a shift of engineer roles towards designing, reviewing and connecting. AI takes over routine coding tasks while allowing engineers to concentrate on high-value-add areas such as analysing content, reviewing code and integrating AI models to optimise design.

In Taiwan, major chipmakers like TSMC, MediaTek and United Microelectronics Corporation (UMC) are working with leading universities across the world to develop AI breakthroughs and integrate these innovations into engineering curricula at local universities such as National Taiwan University and National Tsing Hua University. Tsung-Tsong Wu, Minister of the National Science and Technology Council of Taiwan, said, “When ChatGPT came out — this is smart AI — we are considering what Taiwan should do in semiconductors.” Building on a strong foundation, he says that upgrading core academic and research infrastructure, as well as teaching materials, will enable Taiwan to become a leading centre for engineering talent.

Could we borrow the above strategies and apply them to the Malaysian chip industry? According to the Ministry of Higher Education’s Graduate Tracer Study dataset, Malaysia produced 286,299 graduates in 2021, of whom engineering graduates in the electrical and engineering (E&E) field numbered 23,352. According to the Department of Statistics, there are roughly 640,000 jobs in the E&E field. Out of the 23,352 E&E graduates produced in 2021, actually 14,667 were TVET (technical, vocational educational training) diploma holders. Based on the average trend of 5,827 E&E graduates per year from 2010-2021, Malaysian universities are expected to produce 34,962 graduate engineers within six years, a shortfall from the NSS goal of an additional 60,000 engineers by 2030. 

Thus, a realistic way to meet the staffing shortfall is to retrain existing engineers and TVET staff to upgrade to high-value-added roles such as equipment engineer and packaging and IP designers. Part of the problem of this qualitative shift is persistently rooted in the long approval process for curriculum updates in local universities and the lack of a formal internship, funding and training programme, which should be funded by sources such as the Human Resource Development Fund.

Thus far, Wawasan Open University and University Teknologi Malaysia have taken significant steps in offering generative AI courses. But how many young people will register for them? University students enrolled in science, technology, engineering and mathematics (STEM) subjects grew 43% on average between 2020 and 2023, but a substantial portion of these graduates are working in unrelated fields. Despite the growing interest in technical disciplines, a more coordinated effort between universities and the industry is needed to transform this broad interest into specific AI programme enrollment. This means that the government must step in to accelerate engineering and STEM skills if the NSS is to be achieved.

We believe that in this AI and technology age, there will be no limit to the demand for human talent and re-skilling in a life-long cycle of learning and updating. Hence, we should not worry about over-investment in human talent. The secret is to get the private sector to upgrade market-led specialist skills, with some funding from the government and help from the universities. We all need to change fast to meet the technological and digital challenge.

Tan Sri Andrew Sheng writes on Asian global issues. Loh Peixin is a research associate at the George Town Institute of Open and Advanced Studies, Wawasan Open University. The authors are engaged in a major study of the tech industry in Penang.

Save by subscribing to us for your print and/or digital copy.

P/S: The Edge is also available on Apple's App Store and Android's Google Play.